Skip to main content

FlyMyAI Headless Cloud - platform log

Current version: 1.36.0 (2026-11-10). Newest first. What the platform does today is in SKILL.md; every entry is also at changelog.json. Connected assistants can ask the Agents MCP tool whats_new.

1.36.0 - Long generations as tasks: run_model answers at once, get_model_result reads the result

2026-11-10

  • modelsrun_model no longer holds the call for a whole video. A generation that takes longer than wait_ms (25 seconds unless you set it) answers status "running" with a prediction_id while it goes on, and get_model_result reads the result when it is ready. Reading is free and safe to repeat.
    Use: run_model, get_model_result - docs
    Try: “Generate a 5-second video of a paper plane flying through a neon city and show it when it is ready.”
  • mcpAn assistant can start several generations and collect them as they finish. wait_ms sets how long run_model waits before it hands the generation back; timeout_ms without wait_ms keeps the older call that waits for the result.
    Use: run_model, get_model_result - docs
    Try: “Start three image variations of my logo at once and show each one as it finishes.”

1.35.4 - Read the Hacker News front-page Atom feed

2026-11-10

  • toolsRead the current Hacker News front-page Atom feed as complete XML with your agent, without a provider account. Extract story links and feed metadata from the returned feed_xml.
    Use: search_tools, get_tool_schemas, execute_tool - docs
    Try: “Fetch the Hacker News front-page Atom feed and list its story links.”

1.35.3 - Read and search Hacker News without a provider account

2026-11-10

  • toolsConnect Hacker News without a provider account to read stories, comments and public profiles, browse feeds, and search discussions with filters.
    Use: search_tools, get_tool_schemas, execute_tool - docs
    Try: “Show me Hacker News front-page stories with at least 40 points.”

1.35.2 - ElevenLabs music, multi-voice dialogue and Eleven v4 voices

2026-11-09

  • toolsElevenLabs composes music: compose_music writes a song or an instrumental from a prompt or from a composition plan you can edit, and video_to_music scores background music to your videos. Eleven Music costs $0.15 per minute of music.
    Use: compose_music, music_composition_plan, video_to_music - docs
    Try: “Compose a 30-second upbeat instrumental for my product video with ElevenLabs.”
  • toolstext_to_dialogue voices a conversation between up to 10 ElevenLabs voices in one audio file and, on request, says when each word is spoken and by whom.
    Use: text_to_dialogue, get_voices - docs
    Try: “Make a 20-second podcast intro where two hosts greet each other, with ElevenLabs dialogue.”
  • toolsElevenLabs speech runs on Eleven v4 and v4 Turbo and can return each word's start and end for subtitles, at ElevenLabs' $0.08 per 1,000 characters ($0.04 on v4 Turbo and Flash).
    Use: text_to_speech - docs
    Try: “Read my product announcement aloud with ElevenLabs v4 and give me subtitle timings.”
  • toolsforced_alignment times a transcript you already have against its audio, word by word, and remix_voice changes the accent, age or tone of a voice you designed.
    Use: forced_alignment, remix_voice, create_voice_from_preview - docs
    Try: “Time this transcript against my podcast audio so I can make subtitles.”

1.35.1 - JSON answers under your schema from chat models

2026-11-09

  • apiThe OpenAI-compatible API takes response_format: with json_schema a chat model answers JSON under your schema, with json_object any JSON object. It works on Gemini 2.5 Flash, 3.1 Flash Lite and 3.1 Pro, Grok 4.7, DeepSeek V4.1 Flash and Kimi K3.1 Pro.
    Use: /api/v1/{owner}/{model}/openai/v1/chat/completions - docs
    Try: “Show me how to get invoice fields as JSON under my schema from Gemini 2.5 Flash with the OpenAI SDK.”

1.35.0 - Stop a helper early; Google results with their own pages

2026-11-08

  • agentsA lead can act before its whole batch of helpers ends: delegate_agent or delegate with wake_on_end=true wakes it as soon as that helper ends, so it uses the result and stops a helper it no longer needs with cancel_subagents(child_id). Without it the batch joins as before.
    Use: run_agent, get_run - docs
    Try: “Start two research helpers on the same question and stop the slower one as soon as the first answers.”
  • toolsDataForSEO's Standard queue: dataforseo_serp_google_organic_standard_post queues up to 100 Google searches in one call ($0.06 for 100 at 10 results) and standard_get reads them free. Every organic result comes with its own page; live results often come as google.com/goto links.
    Use: search_tools, execute_tool - docs
    Try: “Check where my site ranks on Google for these 50 searches through the DataForSEO queue.”
  • agentsupdate_agent that only removes a tool, server, resource set or skill saves no new version: the removal already applies to every version, so there is nothing to test or make live.
    Use: update_agent - docs
    Try: “Take Gmail away from my scheduled agent.”

1.34.0 - A frozen version keeps its model and tools

2026-11-08

  • agentsA frozen version keeps the model, effort, web search, tools and skills it was saved with, so schedules and triggers run what you tested. On an agent with a live version, update_agent answers new_version, not live: test it with run_frozen, then set_live_version.
    Use: update_agent, run_frozen, set_live_version, get_compilation - docs
    Try: “Move my scheduled agent to a cheaper model, test the new version, then make it live.”
  • agentsA lead starts the agents its instruction names by id with no subagent_limits.agents list; a list you give stays the whole list. There is still no count limit.
    Use: create_agent, run_agent, list_subagents - docs
    Try: “Create a lead that hands each part of my report to the helper agents I name.”
  • templatesPromptLinks 1.3.2 shows each existing article's Google rank again: Google's results now come as redirect links, and the research agent matches them to their pages without following them.
    Use: projects_templates, projects_copy - docs
    Try: “Copy PromptLinks and find where my site can earn links.”

1.33.1 - Instagram Trend Watcher reports in your language

2026-11-08

  • templatesInstagram Trend Watcher 1.3 writes its sheet tabs, Telegram reports and Gemini notes in English, or in Russian when you write in Russian. Point it at a sheet of your own and the tabs you already use there stay as they are.
    Use: projects_templates, projects_copy - docs
    Try: “Copy the Instagram Trend Watcher template and set it up for natgeo and mkbhd.”

1.33.0 - Agents v1: a named engine and helpers without a count

2026-11-07

  • agentsEvery agent, frozen version and run names its engine, flymy/agent@1: the runtime as it is today. A frozen version keeps the engine it was frozen on and a copy keeps its source's, so a new engine can ship next to it without moving what runs.
    Use: get_agent, list_agent_versions, get_run - docs
    Try: “Which engine do my agent and its live version run on?”
  • agentsA lead can start any of your agents as helpers, with no count limit by default. Each helper's result lists the tools it called, list_subagents shows the tools every agent always has, and a brief over 16 KB is refused with how to pass large inputs by reference.
    Use: run_agent, list_agents, get_team_run - docs
    Try: “Run my lead with five helpers at once and show which tools each one called.”
  • templatesInstagram Trend Watcher 1.2 takes the demo list on «go» and sets up its own two helpers on GPT-6 Luna. SEO / GEO 1.2.1 reads the real site of every Google result again; Website QA 1.2.1 saves every screenshot, so its visual check reads all pages.
    Use: projects_templates, projects_copy - docs
    Try: “Copy the SEO / GEO template and check my site for three queries.”

1.32.5 - Cost options when you freeze an agent

2026-11-06

  • agentsBefore you freeze, get_execution_price with cost_options shows what each run of the frozen version will cost, cheaper models of the same family priced on the same tokens with the update_agent call for each, and what made the run expensive. freeze_agent answers the same.
    Use: get_execution_price, freeze_agent, update_agent, run_frozen - docs
    Try: “Show me the cost options of my agent's last run before we freeze it.”

1.32.4 - Know what a frozen agent will cost before you freeze it

2026-11-06

  • agentsBefore you freeze, assistants check the accepted run's price: every run of the frozen version costs about as much. When it is high, they offer to freeze it as is or a cheaper path that keeps the result complete, such as a cheaper model, tried on the same inputs, and you pick.
    Use: get_execution_price, list_agent_models, update_agent, run_agent, freeze_agent - docs
    Try: “Before you freeze my agent's last run, tell me what each run will cost and whether a cheaper model gives the same result.”

1.32.3 - Connect, share and manage your connections without leaving the Lab

2026-11-06

  • appIn the Lab, Connect sets up a service right there: its keys or the provider's sign-in, and your own MCP server by URL. Manage on a connection renames, pauses, shares or deletes it and adds another account; your own MCP server can be edited, reconnected, paused or deleted.
  • mcpThe Lab's MCP page has the Shared by me, Teams and Agent access tabs: who can use each of your connections, your teams, and which agents may use which connection. Connections shared with you are listed in My MCPs.

1.32.2 - New models: GPT Image 2.6 and Gemini Omni 1.2 Flash

2026-11-06

  • modelsGPT Image 2.6 by OpenAI makes and edits images in two tiers: Flare, the default, and the slower Sunburst for the finest detail. Each tier has a text-to-image and an edit endpoint. An image costs $0.009 to $0.188 by quality and size, $0.032 at the defaults (medium quality, 1024x1024).
    Use: recommend_model, get_pricing, run_model, /api/v1/flymyai/gpt-image-2-6/predict, /api/v1/flymyai/gpt-image-2-6_edit/predict, /api/v1/flymyai/gpt-image-2-6-sunburst/predict, /api/v1/flymyai/gpt-image-2-6-sunburst_edit/predict - docs
    Try: “Draw a watercolor cat reading in a cozy bookshop with GPT Image 2.6, then give it a tiny red scarf with the edit model.”
  • modelsGemini Omni 1.2 Flash by Google makes 3-15 second videos with sound - dialogue, music and effects - up to 4K from a prompt, frames or reference images and clips, and edits or extends a video. A second costs $0.024 at 360p to $0.209 at 4K plus tokens; a 5-second 720p clip costs about $0.36.
    Use: recommend_model, get_pricing, run_model, /api/v1/flymyai/gemini-omni-1_2-flash-t2v/predict/async/, /api/v1/flymyai/gemini-omni-1_2-flash-i2v/predict/async/, /api/v1/flymyai/gemini-omni-1_2-flash-ref2v/predict/async/, /api/v1/flymyai/gemini-omni-1_2-flash-edit/predict/async/, /api/v1/flymyai/gemini-omni-1_2-flash-extend/predict/async/ - docs
    Try: “Make a 5-second 720p video with sound of a street drummer playing in the rain at night with Gemini Omni 1.2 Flash.”

1.32.1 - No default limits on a run's turns, webhook starts or what page visitors spend

2026-11-05

  • agentsA run has no default limit on its turns: it works until the task is done, you cancel it or a loop guard stops a repeat. Research helpers started with delegate have no cap on paid web calls.
    Use: run_agent, run_frozen - docs
    Try: “Research every AI agent platform launched this year and keep going until the table is complete.”
  • toolsA webhook or event trigger has no hourly run limit unless you set rate_limit_per_hour; it pauses itself only when it passes a limit you set.
    Use: triggers_v1_create, triggers_v1_update, triggers_v1_subscribe - docs
    Try: “Start my lead agent on every webhook from my CRM, with no hourly limit.”
  • appPaying for the visitors of a published page or a project's server has no default daily cap: their calls run on your balance until you set a cap per day or per visitor.
    Try: “Publish my demo page and pay for its visitors' runs with a $10 daily cap.”

1.32.0 - Plans are monthly subscriptions, and every month's credits go to your balance

2026-11-05

  • billingSubscribing to a plan in Billing starts a monthly subscription paid by card. Each paid month adds the plan's credits to your balance; they never expire and pay for agents, models and tools. Cancel, resume or switch plans in Billing.

1.31.4 - No spending caps by default: projects and subagents run until you set a limit

2026-11-05

  • billingA new project has no daily, weekly or monthly cap. Set one when you want it in the project's Limits or with projects_set_budget; sending "" for every period removes the budget again.
    Use: projects_create, projects_set_budget - docs
    Try: “Set a $20 daily limit on my newest project, then show me what it spent today.”
  • agentsA run's subagents have no USD cap, count limit or step limit by default, so a long task with many helpers runs to the end. Three run at once and the rest wait their turn; subagent_limits sets a cap, a count or up to 8 at once.
    Use: run_agent, run_frozen - docs
    Try: “Compare the ten biggest AI agent platforms, one helper per platform, and give me one table.”
  • appIn the Lab chat the master says which helper it is waiting for and for how long, writes a short line before each long step, and a used-up project limit shows what was spent with a button to raise it.
    Try: “Have the master research five competitors of my site with one helper each and report as it goes.”

1.31.3 - See what starts a project on events: triggers on the project page and in its diagram

2026-11-05

  • appA project shows its Granola, Telegram Bot and webhook triggers beside its schedules: the status says what starts it, a Triggers section has each one's state, counts and Pause, Activate and Try again, and the diagram draws them. An event's run no longer reads as scheduled.
    Try: “Open my project and tell me what starts it and whether its triggers work.”
  • mcpprojects_get returns a project's triggers with their state and reason, and draws them in the diagram. A fleet counts working, paused and failing triggers; projects_errors lists events that started nothing and sources a provider turned off.
    Use: projects_get, projects_errors, projects_list - docs
    Try: “List my projects whose triggers need attention and why.”

1.31.2 - Claude Haiku 5.5 for agents, files and sign-ins right in the Lab chat

2026-11-05

  • agentsPick Claude Haiku 5.5 as an agent's model: in the model picker, as model claude-haiku-5-5 on the REST API, or in create_agent and update_agent on the Agents MCP. A call costs $0.10 input, $0.01 cached input and $0.50 output per 1M tokens, five times that when its prompt passes 100,000 tokens.
    Use: create_agent, update_agent, list_agent_models - docs
    Try: “Copy my news digest agent onto Claude Haiku 5.5, run it once and compare the price with its last run.”
  • agentsA subagent a lead creates or copies runs on the lead's own model. It takes another model only when you or the lead's instructions name it.
    Try: “Tell my research agent that its source-checking subagent runs on Claude Haiku 5.5.”
  • appIn the Lab, the + menu attaches files to a message: the project's master has them before it reads the message, and a new project gets them right after its brief.
  • appWhen a Lab agent asks for a service, its Connect button opens a window over the chat for keys, a phone code or the provider's sign-in. A service you already connected drops its button.

1.31.1 - Claude Sonnet 5.5 as an agent model, and PromptLinks runs on it

2026-11-05

  • agentsPick Claude Sonnet 5.5 as an agent's model: in the model picker, as model claude-sonnet-5-5 on the REST API, or in create_agent and update_agent on the Agents MCP. Reasoning goes from low to xhigh, and each model call costs $2 input, $0.10 cached input and $10 output per 1M tokens.
    Use: create_agent, update_agent, list_agent_models - docs
    Try: “Move my research agent to Claude Sonnet 5.5, run it once and compare the price with its last run.”
  • templatesPromptLinks 1.3: the lead runs on Claude Sonnet 5.5. A new copy takes this model; a copy you already made keeps the model it has until you change it on the agent and open a new chat.
    Use: projects_copy, append_message, get_execution_price - docs
    Try: “Copy the PromptLinks template, send it example.com and tell me what the run cost.”

1.31.0 - Start agents from Granola and Telegram Bot events, the whole Granola API

2026-11-05

  • agentsSubscribe a frozen agent to your Granola or Telegram Bot connection and it starts on each event: a note generated, edited or shared, or any bot update, with chat and sender filters. FlyMyAI registers the provider webhook and keeps its secret.
    Use: triggers_v1_catalog, triggers_v1_subscribe, triggers_v1_get, triggers_v1_activate - docs
    Try: “Start my meeting agent whenever Granola generates a new note and show me the subscription's status.”
  • toolsThe Granola connector covers the whole public API: note search, transcripts of any length, folders, spaces, audit events, webhook endpoints and legal holds.
    Use: granola_search_notes, granola_get_transcript, granola_list_folders, granola_list_webhook_endpoints - docs
    Try: “Find my Granola notes from last week with anyone at acme.com and list their action items.”
  • agentsA webhook trigger with a whsec_ secret accepts any Standard Webhooks sender as it is, including a webhook made in Granola's app.
    Use: triggers_v1_create, triggers_v1_update - docs
  • toolsTelegram Bot agents can see where a bot's updates go and its last delivery error, and set or remove a webhook of their own.
    Use: get_webhook_info, set_webhook, delete_webhook - docs
    Try: “Tell me where my Telegram bot's updates go and whether its last delivery failed.”

1.30.3 - Freeze a whole project in one call: subagents first, then the lead

2026-11-04

  • agentsFreezing a lead now freezes its subagents first: each one without a current version is frozen from its own run of the subtask, on its own model, then the lead compiles and pins their versions. One freeze or schedule of the lead saves the whole project.
    Use: freeze_agent, get_compilation, schedule_agent - docs
    Try: “Freeze my lead agent's last run together with its helpers and tell me which helper versions it pinned.”

1.30.2 - Provider prices with 0% markup

2026-11-04

  • billingCalls on FlyMyAI's own provider accounts now cost exactly what the provider charges, with 0% markup: DataForSEO, Firecrawl, web search engines, Browser Use and Gemini vision, never less than the call's compute time. Rented servers cost the provider's hourly price.
    Use: list_web_search_engines, get_vision_model, compute_catalog - docs
    Try: “List the web search engines you can use and tell me what one search costs on each.”

1.30.1 - Search Console: filters, fresh data, URL Inspection and sitemap submission

2026-11-03

  • toolsGoogle Search Console reports filter by page, query, country or device, include fresh data from the last days and cover image, video, news and Discover. Agents can also check a URL's index status and read or resubmit sitemaps; earlier connections reconnect once to submit.
    Use: search_tools, get_tool_schemas, execute_tool - docs
    Try: “Resubmit example.com/sitemap.xml and tell me why example.com/blog/post is not indexed yet.”

1.30.0 - FlyMyAI as a Claude Code plugin, smarter tool search, top-up links and feedback

2026-11-03

  • mcpInstall FlyMyAI in Claude Code as one plugin: the Agents MCP 0.7.0-rc.69 with sign-in and the FlyMyAI skill. The guide at flymy.ai/skill.md is also a packaged skill you can save to your skills folder.
    Use: whats_new - docs
    Try: “Install the FlyMyAI plugin in Claude Code and show me three things I can do with it.”
  • toolsTool search says for every match whether it runs right away, what a call costs and how well it worked over the last 7 days, so your agent picks the ready, healthy option.
    Use: search_tools - docs
    Try: “Find a tool that gets Amazon product prices and tell me which one runs now and what it costs.”
  • billingWhen the balance is too low for an agent run, a tool call, a project or compute, the answer says how much is missing and links the finances page, and create_top_up_link makes a checkout link that adds whole dollars.
    Use: create_top_up_link - docs
    Try: “Make me a link to add 20 dollars to my FlyMyAI balance.”
  • mcpYour assistant can send the FlyMyAI team feedback over the Agents MCP, the API or the SDK when a tool fails or is missing, a result looks wrong or the docs are off. The team reads every message.
    Use: send_feedback - docs
    Try: “Tell FlyMyAI that the last tool call returned the wrong data.”
  • sdkPython SDK 1.2.0rc25 makes a top-up link with client.billing.create_top_up_link and sends feedback with client.feedback.send, on both sync and async clients.
    Use: client.billing, client.feedback - docs
    Try: “Write a Python example that makes a 20 dollar top-up link and sends FlyMyAI feedback about a failed run.”

1.29.1 - Find an MCP and start a browser QA project from your assistant

2026-11-03

  • mcpSearch the MCP directory by name or action and filter it by category. Connector pages list their actions and include a setup request for Claude Code, Codex, Cursor or another assistant.
    Try: “Find Telegram in the FlyMy.AI MCP directory and show me its available actions and setup request.”
  • templatesCopy the Responsive Browser Lab to check a website at desktop and mobile viewport sizes. The BrowserStack Device Lab includes a setup brief for connecting your own BrowserStack account before device tests.
    Use: projects_get, projects_copy - docs
    Try: “Show me the Responsive Browser Lab project and its setup brief before we choose a website and budget.”

1.29.0 - Telegram: create and run groups, channels and forum topics from your account

2026-11-03

  • toolsThe personal Telegram connector creates a group, supergroup, forum or channel with the people you name, adds and removes members, makes invite links for people whose privacy blocks adding them, joins and leaves chats, and edits the title, description and photo.
    Use: telegram_create_group, telegram_add_chat_members, telegram_create_invite_link, telegram_join_chat, telegram_edit_chat - docs
    Try: “Create a Telegram group Launch Crew with two of my contacts and post a welcome message there.”
  • toolsRun a Telegram group or channel: promote admins with chosen rights, ban or mute members, set default permissions, approve join requests, create forum topics and post into them, and read the admin log and channel stats.
    Use: telegram_edit_admin, telegram_edit_permissions, telegram_handle_join_requests, telegram_create_forum_topic, telegram_get_admin_log, telegram_get_channel_stats - docs
    Try: “Add a topic Weekly updates to my Telegram forum and post this week's agenda in it.”
  • toolsTelegram messages from chat: schedule a message, file or forward for later and cancel it, send albums, polls and quizzes, locations and contacts, forward as a copy, comment under channel posts, and read history by sender, dates, topic or kind.
    Use: telegram_send_message, telegram_send_album, telegram_send_poll, telegram_list_scheduled_messages, telegram_get_messages - docs
    Try: “Schedule a Telegram poll in my team chat for Friday 10:00: which demo day works best?”
  • toolsSee how Telegram messages land: views, forwards and reactions, who read a message in a small group, poll results, translations, and voice message transcripts (Telegram Premium or its weekly free tries).
    Use: telegram_get_message_stats, telegram_get_poll_results, telegram_translate_messages, telegram_transcribe_voice - docs
    Try: “Who reacted to my last message in my team's Telegram chat, and with what?”
  • toolsYour Telegram account from chat: add or remove contacts, block users, see chats you share with someone, mute, pin or archive chats, build chat folders, update your name, bio, username and photo, and post or read photo stories.
    Use: telegram_add_contact, telegram_update_dialog, telegram_save_folder, telegram_update_profile, telegram_post_story - docs
    Try: “Put my Telegram chats with clients into a folder named Clients and mute the noisy ones.”

1.28.0 - Save a whole project at once; a copy of a public project comes ready to run

2026-11-03

  • agentsA new project's agent hands every task to a helper, even a one-off one: it reuses a helper that fits before it makes a new one, then collects and delivers the result, and answers questions about the project itself. Earlier projects keep working as before.
    Use: projects_create
    Try: “Create a project that finds one fresh space news story and tell me which helper found it.”
  • sharingA copy of someone else's public project comes ready to run: its saved versions, its files and the author's example results come with it, and its agent opens with a welcome that says what to connect and give. The author's schedule is named, not copied.
    Use: projects_copy
    Try: “Copy the public project at this link into my account and tell me what it needs before its first run.”
  • apiSave a project in one request: its helpers are saved first, then its agent, whose new version runs theirs. The save goes on in the background; read it until it is done.
    Use: /api/v1/agents/projects/{id}/save/, /api/v1/agents/projects/{id}/save/{save_id}/
  • apiMove a saved version of a project's agent onto its helpers' current versions without compiling it again. A project's agents now show their saved version and the one their lead runs.
    Use: /api/v1/agents/compilations/{id}/repin/, /api/v1/agents/projects/{id}/

1.27.4 - PromptLinks 1.2: send your website, get link targets for a mind map of searches

2026-11-03

  • templatesCopy the PromptLinks template and send only your website: it writes a mind map of 50-100 buyer searches, checks each on Google, its AI Overview and the web, and lists PR publishers from your Collaborator catalog and existing articles to ask for a link. No research cap to set.
    Use: projects_templates, projects_copy, append_message, get_run - docs
    Try: “Copy the PromptLinks template and find where example.com can earn links.”

1.27.3 - Make MiniMax H3 videos and run DeepSeek V4.1 Flash 1017 with its native input

2026-11-03

  • modelsMiniMax H3 makes a 5- or 10-second video with sound from your prompt and 1 to 12 of your own reference images and audio clips. Run it from the Python SDK, the Agents MCP or the native model API and download the video; the price depends on the video length.
    Use: recommend_model, get_pricing, run_model, /api/v1/model-proxy/operations/ - docs
    Try: “Make a five-second MiniMax H3 video from the photo and voice clip in my FlyMyAI files.”
  • modelsDeepSeek V4.1 Flash 1017 runs with its native chat input and output, function calls included, and cached input is priced lower. The answer comes in one piece, without streaming. Read its schema and price, then call it from Python, the Agents MCP or the native model API.
    Use: get_pricing, run_model, /api/v1/model-proxy/operations/ - docs
    Try: “Ask DeepSeek V4.1 Flash 1017 to explain this Python function and suggest one test.”

1.27.2 - Send Telegram bot reports as text attachments

2026-11-03

  • toolsTelegram Bot sends a UTF-8 text file and a short caption together. Choose the destination chat explicitly and deliver longer reports as an attachment.
    Use: search_tools, get_tool_schemas, execute_tool - docs
    Try: “Send my latest report as a text file with a short caption through my Telegram bot.”

1.27.1 - OmniHuman 2.0: a picture and a voice track become a talking video

2026-11-03

  • modelsOmniHuman 2.0 by ByteDance turns one picture and an audio track into a video of the person, character or animal in it talking, singing or performing in time with the sound, from 4 seconds to 5 minutes in 720p, for $0.07 per second of video.
    Use: run_model, recommend_model, get_pricing, /api/v1/flymyai/bytedance-omnihuman-2_0/predict/async/ - docs
    Try: “Turn this photo and my voice note into a talking video with OmniHuman 2.0.”

1.27.0 - Read a public project's chats and files, see a project on Preview, delete template copies

2026-11-02

  • sharingSomeone else's public project opens in the app with its master's chat, its helpers' chats and its files, read only, beside its Preview and Copy project. projects_get returns the same chats to an assistant.
    Use: projects_get, projects_copy - docs
    Try: “Open the public project link I paste and sum up what its master did in its chat.”
  • sharingAnyone with a public project's link reads its agents' chats and files, read only, so keep secrets and personal data out of them. A copy still takes no chats or data.
    Use: projects_share - docs
  • sharingMaking a project or an agent public needs a saved version first: Share asks you to save one, and projects_share and set_agent_public answer no_saved_version until then.
    Use: projects_share, set_agent_public, freeze_agent - docs
    Try: “Save a version of my project, then make it public and give me its link.”
  • appA project's panel opens on Preview: what the project does in plain words, its status, and its diagram, which opens full size and updates as helpers, tools and models are added.
    Try: “Open my project's Preview and show its diagram full size.”
  • appProjects copied from templates can be deleted from the ⋯ menu, in the projects list and on the project's page.
    Try: “Open my projects list and delete the template copy I no longer need from its ⋯ menu.”
  • agentsAn agent's preview, the short description people see before they copy it, is written in the owner's language and stays the same while the agent's goal does not change.
    Use: get_agent, update_agent - docs
    Try: “Show my agent's preview and rewrite it in two short sentences.”

1.26.1 - See media and each agent step in project chats, and call saved projects from code

2026-11-02

  • appA project's chat shows the pictures, videos and sounds its agents link to, and a picture opens full size. The project's diagram shows the models and connections it uses, each with its mark and name.
    Try: “Open a project and ask its master for three logo ideas for a coffee shop called Bean There.”
  • appEach stretch of an agent's work in a project chat shows its plan, the tools and code it called, events such as a saved version, and why it stopped: a step limit, a cancel, no balance or a wait for approval.
    Try: “Ask a project's master to compare three coffee shops in Austin and watch its plan and tools as it works.”
  • appRun it in a project shows its webhooks and the code that calls its newest saved version from Python or cURL. Show key puts your API key into that code for 30 seconds.
    Try: “Open Run it in my project and copy the Python call with my key in it.”
  • appImage, video and audio inputs of a model's page in the app take files from your computer as well as links, each with a preview.
    Try: “Open an image editing model and upload a photo from your computer as its reference.”
  • appAn app's secrets are in its project's Connections tab, next to the services its agents use. Enter a value there, never in a chat; a dot on the tab marks a secret the app's servers still need.
    Try: “Open my shop app's Connections tab and see which secrets its server still needs.”
  • appA helper's page in a project shows the tasks its master gave it with their answers, and an answer in JSON reads as a report with its problems, links and facts.

1.26.0 - Choose project effort and manage app secrets

2026-11-01

  • agentsChoose a supported effort level for a new project through REST, remote Agents MCP or Python. Preview before creating; without a brief, creation saves the agent's settings without starting a task.
    Use: list_agent_models, projects_plan, projects_create - docs
    Try: “Preview a private project with a supported model and effort level from the current catalog, without creating it.”
  • appRead app-secret metadata, generate an absent value and delete it through REST, remote Agents MCP or Python. These responses contain metadata, never the stored value; repeated generation keeps an existing value.
    Use: apps_secrets_list, apps_secrets_generate, apps_secrets_delete - docs
    Try: “Show the secret names and status for my app, without changing any values.”

1.25.2 - MCP guides for dashboards, feature flags and code

2026-10-31

  • mcpFind connection steps, listed actions and scoped read examples for Grafana, LaunchDarkly, Sourcegraph and other existing integrations.
    Try: “Show me the Grafana MCP guide, its connection steps and three read-only requests I can try.”

1.25.1 - More setup guides for existing MCP integrations

2026-10-31

  • mcpExplore 64 additional guides for existing integrations, including Confluence, Microsoft Teams, Salesforce and WordPress. Each guide has connection steps, listed actions and bounded read requests.
    Try: “Show me the Confluence MCP guide, its connection steps and three read-only requests I can try.”

1.25.0 - Start projects with a task and browse saved agent versions

2026-10-31

  • apiStart a project with a first task and chosen model through Agents MCP, REST or Python. Review its plan, create it once, then continue the returned conversation. Repeating the same request and creation key returns that project and conversation.
    Use: projects_plan, projects_create - docs
    Try: “Plan a private writing project whose first task drafts a short welcome message, and show its model and budget.”
  • agentsBrowse an agent's saved versions through Agents MCP, REST or Python. See its live version and schedule state, then open the instructions for the version you choose.
    Use: list_agent_versions, get_compilation - docs
    Try: “Show the saved versions of my agent and open the instructions for the version I choose.”
  • sdkPython supports project title, budget and deletion methods on both sync and async clients. Use the same idempotency key when retrying an edit, and copy projects directly from their current share links.
    Try: “Show a Python example that renames a project and changes its daily budget.”

1.24.4 - Find more MCP connection guides and read requests

2026-10-31

  • mcpOpen connection guides for Gmail, Google Drive, Slack, GitLab and more services in the MCP catalog. Each guide explains the account or key you need, lists supported actions and offers scoped read requests to try with your own connection.
    Try: “Show me how to connect GitLab through FlyMy.AI and which read-only requests I can try.”

1.24.3 - Find setup guides and requests in the MCP catalog

2026-10-31

  • mcpBrowse setup guides, available actions and example requests for Firecrawl, Attio, Google Search Console, Ahrefs, Semrush, HubSpot, Linear, Resend, Stripe, Asana, Trello and Supabase in the MCP catalog.
    Try: “Show me how to connect Attio through FlyMy.AI and what I can ask it to do.”

1.24.2 - Retry a Python chat follow-up with the same key

2026-10-30

  • sdkPython SDK: runs.append_message accepts idempotency_key on both sync and async clients. Reuse the key after a lost response to continue the chat once; the same request returns its first acknowledgement for 30 days. Give each intentional follow-up a new key.
    Try: “Show me a Python example that continues a chat safely after a lost response.”

1.24.1 - Choose how large the FlyMy.AI interface looks

2026-10-30

  • appIn Account & Team, open Display and choose Compact (90%), Default or Large (110%) for text and spacing. The choice stays in this browser across Lab and the classic interface, where Display is in Profile. Default follows your browser's font size.
    Try: “Open Display and make the FlyMy.AI interface more compact in this browser.”

1.24.0 - Copy FlyMy.AI's official templates

2026-10-30

  • templatesprojects_templates also lists FlyMy.AI's official templates that can be copied now, each with its version and how many agents a copy brings. Copy one with projects_copy and its project: it opens with a welcome chat that asks what to set up.
    Use: projects_templates, projects_copy, GET /api/v1/agents/projects/templates/ - docs
    Try: “Show me FlyMy.AI's official templates and copy the SEO / GEO one for example.com.”

1.23.0 - Project limits hold schedules and helpers; change a budget, rename or delete by API

2026-10-30

  • agentsA scheduled agent waits instead of failing while its project is stopped or a spending limit is used up. The skipped run shows in its runs, you get one email per limit period, and the schedule runs again once the limit resets, you raise it or you start the project.
    Use: projects_stop, projects_start, list_compilations - docs
    Try: “Stop my trendwatch project for the weekend and tell me when its scheduled runs will start again.”
  • billingHelper agents that a project's agent creates spend under the project's budget on their own schedules and chats too, so the project's caps hold them and stopping the project stops them.
    Use: projects_get, projects_stop - docs
    Try: “Show what my project's helper agents spent today against the project's daily limit.”
  • apiPATCH /api/v1/agents/projects/<id>/budget/ (MCP projects_set_budget) sets a project's daily, weekly or monthly limit, and null removes one. It releases the project's config with only that change, and what was spent stays.
    Use: projects_set_budget, PATCH /api/v1/agents/projects/<id>/budget/ - docs
    Try: “Raise the daily limit of my trendwatch project to $10 and drop its weekly limit.”
  • apiA project has a title it lists under, apart from its id: PATCH /api/v1/agents/projects/<id>/ (MCP projects_rename) renames it. DELETE (projects_delete, after your OK) deletes it: one that never ran is removed, one with history stops and leaves your lists while its runs and billing history stay.
    Use: projects_rename, projects_delete, PATCH /api/v1/agents/projects/<id>/, DELETE /api/v1/agents/projects/<id>/ - docs
    Try: “Rename my privet-naydi project to Weekly AI news.”
  • apiFreezing an agent takes an optional Idempotency-Key: a retry with the same key within 30 days returns the first version instead of paying for a second freeze, and the same request sent while that freeze still compiles gets it too.
    Use: freeze_agent, POST /api/v1/agents/compilations/freeze-instruction/{execution_id}/, POST /api/v1/agents/compilations/schedule/{execution_id}/
    Try: “Freeze my last chat as a new version, and if the call times out, retry it safely with the same key.”
  • apiContinuing a chat takes an optional Idempotency-Key (operation_key on append_message): a retry with the same key and message within 30 days returns the first answer with Idempotent-Replayed: true instead of adding the message again and paying for a second run. Another message under it is refused.
    Use: append_message, POST /api/v1/agents/executions/{execution_id}/append-message/ - docs
    Try: “Ask my last chat to make its summary shorter, and if the call times out, retry it safely with the same key.”
  • billingBilling and the Upgrade window in the app name the account a top-up or a plan is for. On your personal account, if you are an Owner or Admin of a team, a button switches to paying for the team: money on the personal balance never reaches a team.
  • appOn a project's page, the menu next to Stop renames the project, sets its daily, weekly and monthly limits, and deletes it once you type its name. A project that never ran is removed for good; one with history is archived, and its runs and billing stay.

1.22.0 - Frozen project agents run code on your own runs

2026-10-29

  • agentsA frozen agent in your flymy.yaml project, the kind a schedule runs, now has the code sandbox on your own runs: its schedule, the project's Start and run_frozen. Runs started by page visitors, triggers or other inbound channels stay without it.
    Use: apps_apply, run_frozen - docs
    Try: “Schedule my project agent to compute this week's top models with code and save the result to the project's storage.”

1.21.0 - Project agents use any tool from the catalog

2026-10-29

  • toolsAn agent in your flymy.yaml project can use any tool from the FlyMyAI catalog on your own connection, as in the chat: Slack, Notion, SerpAPI, Apify, Hugging Face and the rest. Name its slug in a ToolAccess with mcp; the plan lists each tool you still need to connect, with its link.
    Use: apps_plan, apps_apply, apps_catalog - docs
    Try: “Add Slack and Notion tools to my project agent in flymy.yaml and show me the plan.”

1.20.0 - Run external chat models with their native input and output

2026-10-28

  • apiA new REST API runs external chat models with their native chat completions input and returns their native output. Each model has a versioned contract with its schemas and per-token price: the maximum charge is reserved when a call starts, and the exact charge is taken once when it ends.
    Use: /api/v1/model-proxy/catalog/, /api/v1/model-proxy/operations/ - docs

1.19.0 - Agents search the web with their model's own search, or the engine you pick

2026-10-28

  • agentsEvery agent has web search built in, on its model's own search by default: OpenAI web search for GPT, Google Search for Gemini, Claude web search for Claude, Grok web search for Grok, and Exa for open-weight models. A search returns the pages found and, from a model's search, an answer with sources.
    Use: create_agent, run_agent, list_web_search_engines
    Try: “Find this week's three biggest AI model releases and cite where each was announced.”
  • agentsPick the engine per agent with web_search on create_agent or update_agent, or the Search picker in the agent chat: auto, an engine such as exa, perplexity, tavily or google_serp, or off. Name an engine in chat to compare engines on one question. Each search is billed at its engine's price.
    Use: update_agent, list_web_search_engines
    Try: “Search who leads the 2026 Formula 1 championship with Exa and with Perplexity, and compare the answers.”

1.18.3 - Firecrawl without your own key

2026-10-28

  • toolsAgents, direct tool calls and a project's flymy.yaml can scrape, map, search and crawl sites with Firecrawl on the platform's key, without connecting your own: a call is billed by the Firecrawl credits it takes and held by the app's budget. A Firecrawl key you connected is still used first.
    Use: execute_tool, add_tool, apps_plan
    Try: “Scrape our pricing page and the three competitor pricing pages with Firecrawl and compare the plans in a table.”

1.18.2 - Copies open as projects; public code blocks; agent code keeps state

2026-10-27

  • sharingClone or copy an agent, a project or a template and the copy opens as its project page with a Start card that opens its setup chat; Connect buttons in that chat show Connected as soon as the service is linked.
    Use: projects_copy, copy_agent, projects_get
    Try: “Copy the Website QA template and open its project page so I can start it.”
  • sharingPublish an agent that runs code blocks: everyone who copies it gets their own copy of each block, the same code under their account. A single agent now also shares and copies as a one-agent project.
    Use: set_agent_public, projects_share, projects_get, copy_agent
    Try: “Make my agent with code blocks public and share its project link.”
  • agentsCode an agent runs in its sandbox can keep state for its next runs with flymy.save(key, value, scope="task", namespace=..., memory_type=..., description=...), the same memory put_memory_item writes, without copying large values through the chat.
    Use: put_memory_item, run_agent
    Try: “Make my daily digest agent keep yesterday's list in memory from its script and send me only what is new.”
  • appAgents in a project's flymy.yaml can search the web with Tavily on the platform's key, billed per call and held by the app's budget, and read sites with Firecrawl through your own Firecrawl connection.
    Use: apps_catalog, apps_plan, apps_apply
    Try: “Add a web research agent with Tavily to my project and show me the plan.”

1.18.1 - Website QA template: a visual check of every page

2026-10-27

  • templatesThe Website QA template now also looks at each page: the cloud browser saves up to three screens of every page on mobile and desktop, and Gemini 3.8 Flash lists what a visitor sees broken - overlapping or cut-off text, words run together, broken images, empty areas - with screenshots in the report.
    Use: projects_copy, copy_agent - docs
    Try: “Clone the Website QA template and check https://example.com on mobile and desktop with screenshots.”

1.18.0 - Teams: invite by email or username, Admin and Member roles

2026-10-27

  • sharingInvite people to a team by email, with or without a FlyMyAI account: they sign up with that address and join when they accept, no extra approval. Pending invitations sit in the team's member list with Resend and Revoke; a username is added right away.
    Use: list_team_members, invite_team_member, remove_team_member
    Try: “Who is in my team, and who is still invited?”
  • sharingTeams have two roles: Admins invite and remove people and change roles, Members work as the team. Everyone in the team uses its credits in the app and from Claude or ChatGPT connected as the team. Change a role in the member list or with set_team_member_role.
    Use: set_team_member_role, list_team_members
  • sharingMCP connections are shared until revoked by default, with one person or a whole team; pick 1, 7, 14 or 30 days only when you want an end date. Everyone in a team can use the connections shared with it.
    Use: invite_mcp_user, list_mcp_shares

1.17.6 - PromptLinks template; copy before services are connected

2026-10-27

  • templatesClone the PromptLinks official template: for your site, brand and topic two research agents find publishers for new PR placements in your Collaborator catalog and existing articles worth a link, on one private page with two CSV tables. Publisher research needs Collaborator catalog API access.
    Use: projects_copy, copy_agent - docs
    Try: “Clone the PromptLinks template and find link opportunities for example.com about project management.”
  • sharingCopy an agent, a project or an official template before its services are connected: the copy is made at once and its setup chat sends every missing connection link in one message, then waits until they are connected. Over the API add connect_later: true; Agents MCP copies do it for you.
    Use: projects_copy, copy_agent
    Try: “Copy the PromptLinks template before I connect Collaborator and send me the link to connect it.”

1.17.5 - SEO / GEO template: search visibility and AI Overview citations

2026-10-27

  • templatesClone the SEO / GEO Research Agent official template: for a domain and 3-10 target queries it reads Google organic results and AI Overview references, crawls up to 60 public pages and returns up to 10 prioritized actions with evidence on a private report page. Nothing to connect.
    Use: projects_copy, copy_agent - docs
    Try: “Clone the SEO / GEO template and check example.com for three queries about my product.”

1.17.4 - Website QA template: check a site on mobile and desktop

2026-10-26

  • templatesClone the Website QA official template: it checks every link of a site, opens each page in a real cloud browser on mobile and desktop, and publishes a private report page with broken links, pages wider than the screen, broken images and tiny tap targets. Nothing to connect.
    Use: projects_copy, copy_agent - docs
    Try: “Clone the Website QA template and check https://example.com on mobile and desktop.”

1.17.3 - Agent previews in the Python SDK

2026-10-25

  • sdkPython SDK 1.2.0rc18 reads an agent's preview and preview_pinned and sets the preview with agents.create(..., preview=...) and agents.update(..., preview=...).
    Try: “With the flymyai 1.2.0rc18 SDK, set a one-line preview on one of my agents and read it back.”

1.17.2 - Use-case guides for building and shipping AI agents

2026-10-24

  • docsBrowse task-based guides for creating, scheduling, embedding and deploying AI agents. Each guide explains the workflow, setup, output and cost considerations, with links to the current documentation.
    Try: “Show me the FlyMyAI guide for adding an AI agent to my SaaS.”

1.17.1 - A short preview on every agent people copy

2026-10-24

  • sharingEvery agent gets a preview - one or two sentences on what it does - that FlyMyAI writes each time a version goes live. People see it where they copy the agent: on its public page and, for a lead, as its project's description.
    Use: get_agent, projects_get - docs
    Try: “Freeze my last good agent run and tell me the preview people will see when they copy it.”
  • sharingWrite your own preview in the agent's Share window or with update_agent (preview, up to 280 characters); it stays through later freezes until you clear it.
    Use: update_agent - docs
    Try: “Set the preview of my newest agent to one clear sentence about what it does, then show it to me.”

1.17.0 - Clone official agent templates with a guided setup

2026-10-24

  • templatesClone an official agent template in one click. The copy brings the lead agent with its guide, scripts and a demo input - never the author's chats, helpers, data or connections - and opens a welcome chat that asks everything once, then sets up and schedules the helpers.
    Use: projects_copy, copy_agent - docs
    Try: “Clone the Instagram Trend Watcher template and set it up for @natgeo and @nasa with a daily report.”
  • templatesTemplate pages show the project diagram, every agent's tools with unit prices, how the template runs on a schedule and its measured setup cost.
    Try: “Open flymy.ai/templates and tell me what a day of the Instagram Trend Watcher costs for 20 accounts.”
  • appAgents without a project are listed on the Projects page; the Agents page now opens Projects.

1.16.7 - Kimi K3.1 Pro as an agent brain and through the API

2026-10-23

  • modelsKimi K3.1 Pro, Moonshot's thinking model for agents and code, answers through /predict, the OpenAI-compatible API with function tools, and run_model, and reads images: $2 input and $10 output per 1M tokens. Set service_tier to priority for its faster Highspeed serving, or turn on private_mode.
    Use: run_model, get_pricing, /api/v1/flymyai/kimi-k3_1-pro/openai/v1/chat/completions, /api/v1/flymyai/kimi-k3_1-pro/predict - docs
    Try: “Ask Kimi K3.1 Pro on the priority tier to write a Python script that renames my photos by the date they were taken.”
  • agentsPick Kimi K3.1 Pro as an agent's brain: in the model picker, as model kimi-k3.1-pro on the REST API, or in create_agent and update_agent on the Agents MCP. Its reasoning goes from low to max, and each model call costs $2 input and $10 output per 1M tokens.
    Use: create_agent, update_agent, list_agent_models - docs
    Try: “Make an agent on Kimi K3.1 Pro that finds this week's three most discussed AI papers and sums up each in two lines.”

1.16.6 - Retry image and video generation without paying twice

2026-10-23

  • mcprun_model takes an optional operation_key for image and video models too. The same key and input return the first job without a second charge; the same key with other input is refused before anything runs. Text models refuse a key.
    Use: run_model - docs
    Try: “Make a flat blue circle icon with an operation key, then send the exact same call again and confirm I paid once.”
  • apiPOST /predict/async/ accepts an Idempotency-Key header. The same key and input return the first prediction_id instead of queueing a second paid run; another input under that key gets 409 idempotency_key_reused.
    Use: /api/v1/{owner}/{model}/predict/async/ - docs
  • modelsModel schemas describe each input's limits, such as allowed durations, file types and minimum video size, in get_model_schema and the model's openapi.json.
    Use: get_model_schema, /api/v1/{owner}/{model}/openapi.json - docs
    Try: “Show me the input limits of Seedance 2.5 reference-to-video before I upload my clips.”

1.16.5 - Plan AI search citations with PromptLinks and CiteAI

2026-10-22

  • templatesUse the PromptLinks starter brief to research SEO and GEO link prospects, compare publishing options and review crawler access and source credibility before choosing placements.
    Try: “Show me the PromptLinks brief and the connections I need to research link prospects.”
  • templatesUse the CiteAI starter brief and downloadable runbook to plan citation placements, prepare outreach and track source-backed results with review checkpoints.
    Try: “Open the CiteAI runbook and help me plan a citation-placement workflow.”
  • mcpRead the Collaborator and DataForSEO setup guides, choose the connections for your research workflow and copy a scoped setup brief before running it.
    Try: “Show me the Collaborator and DataForSEO guides and which connections PromptLinks uses.”

1.16.4 - Investment research workflows with Akta company data

2026-10-20

  • templatesStart investment research from editable Akta workflows: source private companies for a thesis, check growth signals, write a first-pass memo, compare competitors, read customer reviews or find a timely reason to reach a founder.
    Try: “Use Akta to write a first-pass investment memo on Linear with sources, three risks and five founder questions.”

1.16.3 - Find agent templates and connector setup guides

2026-10-19

  • templatesBrowse Instagram trend-watching and SEO/GEO research starter templates. Each page includes a downloadable brief, required connections, inputs, outputs and usage pricing to help you create your own agent.
    Try: “Show me the Instagram trend-watching template and what I need to connect before creating my agent.”
  • mcpUse dedicated Instagram and Telegram guides to choose a connector, understand its access and pricing, and copy a scoped setup request into your assistant.
    Try: “Show me how to connect Telegram to my agent and start with a read-only summary of a chat I choose.”

1.16.2 - Instagram data requests cost $0.01

2026-10-19

  • billingRead public Instagram profiles, posts, reels, comments and search results for $0.01 per billed request with the shared FlyMyAI connection, reduced from $0.02. New agent runs, direct tool calls and Team reservations use this price.
    Try: “Look up the public Instagram profile @instagram and show its follower count.”

1.16.1 - Retry paid vision and decision calls safely

2026-10-19

  • mcpanalyze_media, make_typed_decision and run_model for the vision model or TypeSafe Jev take an optional operation_key. The same key and input return the first answer without a second charge; the same key with other input is refused before anything runs.
    Use: analyze_media, make_typed_decision, run_model - docs
    Try: “Describe this product photo with an operation key, then send the exact same call again and confirm I was charged once.”
  • apiWhen an Idempotency-Key was already used, the 409 answer carries a code: idempotency_key_reused when the key belongs to another request, idempotency_key_in_flight while the same request is still running.
    Use: /api/v1/agents/tasks/{id}/run-loop/, /api/v1/agents/compilations/{id}/run-instruction/ - docs

1.16.0 - Pass complete files and instructions between agents

2026-10-19

  • agentsGive an agent instructions from an uploaded file. Helpers return complete reports by reference, and agent code can stream their contents with size and SHA-256 verification.
    Use: upload_agent_file, create_agent, run_agent, read_run_result - docs
    Try: “Give a helper my instruction file, have it build a report, and pass the complete report to the lead.”
  • apiStart an agent with its first message in the same request. Owner runs can create and schedule requested helpers, with input values that can change for future scheduled runs.
    Use: run_agent, create_agent - docs
    Try: “Set up a helper from my instructions, test it with these inputs, then schedule it for Monday mornings.”

1.15.4 - DeepSeek V4.1 Flash through the API, with a faster priority tier

2026-10-18

  • modelsDeepSeek V4.1 Flash answers with text and images through /predict, the OpenAI-compatible API with function tools, and run_model: $0.30 input and $1.20 output per 1M tokens. Set service_tier to priority for its fastest serving, or private_mode to keep prompts and answers out of FlyMyAI storage.
    Use: run_model, get_pricing, /api/v1/flymyai/deepseek-v4_1-flash/openai/v1/chat/completions, /api/v1/flymyai/deepseek-v4_1-flash/predict - docs
    Try: “Ask DeepSeek V4.1 Flash on the priority tier to write a Python function that checks whether a string is a palindrome.”

1.15.3 - Read what a run used and cost from the Python SDK

2026-10-18

  • sdkThe Python SDK reads what a run used and what it costs: client.runs.prices(run_id) for one run and client.runs.list_prices() for your newest runs or the ones you name, sync and async, from flymyai 1.2.0rc16.
    Use: GET /api/v1/agents/executions/{id}/prices/ - docs
    Try: “How much did my last ten agent runs cost, and which one used the most model tokens?”

1.15.2 - See which lines of a run's cost are already charged

2026-10-18

  • billingEvery tool call and model answer in a run's cost says whether it is settled - already charged - or still the estimate the charge will use, and model answers show their reasoning tokens.
    Use: get_execution_price, list_execution_prices, GET /api/v1/agents/executions/{id}/prices/
    Try: “Is the cost of my last agent run final yet, and how many reasoning tokens did it use?”
  • appThe cost breakdown in the chat header shows reasoning tokens next to tokens out and says how many lines are not charged yet.

1.15.1 - Choose DeepSeek V4.1 Flash 1017 as your agent brain

2026-10-18

  • agentsChoose DeepSeek V4.1 Flash 1017, DeepSeek's October 17 update of V4.1 Flash, in the model picker or from the Agents MCP. It takes the place of V4.1 Flash among the low-cost models for helper agents and subagents; agents already set to V4.1 Flash keep it.
    Use: list_agent_models, create_agent, update_agent, run_agent
    Try: “Move my helper agent to DeepSeek V4.1 Flash 1017 and run it once.”

1.15.0 - Start an agent from a signed webhook

2026-10-16

  • agentsAn agent's live version can start from a signed webhook. Create one in the agent's Run it tab (On an event), the API or the Agents MCP, sign each JSON event with your secret, and each event ID starts at most one run. Pause it, rotate the secret or read recent deliveries at any time.
    Use: triggers_v1_create, triggers_v1_list, triggers_v1_events, triggers_v1_pause, /api/v1/agents/triggers/ - docs
    Try: “Show the webhooks on my agents and which recent deliveries started a run.”
  • sdkThe Python SDK manages webhooks with client.triggers_v1: create, update, pause, activate, delete and read recent deliveries, with an idempotency key on every write.
    Use: client.triggers_v1 - docs
    Try: “Write a Python script that lists my webhooks and prints each recent delivery with its run link.”

1.14.0 - Share a project by link and copy someone's project

2026-10-13

  • sharingA project shares the way an agent does: Share, then Make public & copy link gives https://app.flymy.ai/projects/<id>. Anyone with the link sees its agents, pages and what a copy needs to connect - never its data, conversations, money or connections - and Make private stops new copies.
    Use: projects_share, projects_get - docs
    Try: “Make my project public and give me its link to post.”
  • sharingCopy project brings a public project into your account: its agents with all their subagents, its pages and config, under your own budget. A tool you have not connected is asked for first; then the setup assistant connects what is left - a bot token, inputs, a schedule - in its chat.
    Use: projects_copy - docs
    Try: “Copy https://app.flymy.ai/projects/fleet%3A94a14ca2-bd5c-4de4-bd32-54a3a062b1ed into my account and run it on https://example.com.”

1.13.3 - Agents create agents from your API runs; your Telegram bot names itself

2026-10-13

  • agentsA run you start through the API, the SDK or the Agents MCP can create and edit your agents with create_subagent and update_subagent, as your app chat does, and start them with delegate_agent. Scheduled and frozen runs, subagents and a deployed agent's customers cannot.
    Use: run_agent, create_subagent, update_subagent, delegate_agent - docs
    Try: “Run my agent: create a subagent that sums up a web page in three bullets, run it on https://example.com, and show me its result.”
  • toolstelegram_bot get_me names your bot (its @username) and the chat a message without chat_id goes to. A send to a bot nobody has written to yet says to open the bot in Telegram and press Start.
    Use: telegram_bot.get_me - docs
    Try: “Which Telegram bot is connected to my account, and which chat will it write to?”
  • agentsA failed tool call in a run's steps says why, in one line (reason), for the run's owner.
    Use: get_run - docs
    Try: “Show me why the failed steps of my last agent run failed.”
  • billinglist_agent_models names the cache-write price per 1M tokens for models that charge it, next to input, cached input and output.
    Use: list_agent_models - docs
    Try: “List the agent models with their input, cached input, cache-write and output prices.”
  • appRun now and Try it once on a saved version ask for its required inputs, prefilled from its schedule or its examples.

1.13.2 - Sandbox code runs until it is done

2026-10-12

  • toolsexecute_code, execute_code_with_outputs and run_code_block no longer stop at a default time limit: code runs until it finishes, its run ends, its budget or balance runs out, or the sandbox session ends (up to 30 minutes). Pass timeout_seconds to stop it sooner.
    Use: execute_code, execute_code_with_outputs, run_code_block - docs
    Try: “Run a Python script in the sandbox that works for three minutes, printing a line every 30 seconds, and show me what it printed.”

1.13.1 - Edit artifacts in place; project budgets hold the project's own agent

2026-10-12

  • artifactsChange an artifact without sending whole files: publish_artifact_version takes edits (exact text replacements) and files[].from (copy a file from an artifact you can read), with an optional sha256 check. read_artifact_files finds text, reads line ranges and describes images.
    Use: publish_artifact_version, read_artifact_files - docs
    Try: “Change the stage 1 price in my proposal deck to $30,000 and copy the logo from my other deck.”
  • billingA project's budget now also counts and caps what the project's own agent and the helpers it starts spend, as it does for the agents its flymy.yaml declares. When a cap is reached, the next paid call is refused with the usual budget message.
    Try: “How much have my project's agent and its helpers spent today against the project budget?”
  • agentsIn your own chat an agent can test an edited code block before pinning it: run_code_block with version "draft" runs the current draft, then freeze_code_block pins it. Frozen versions keep running their pinned code.
    Try: “Edit my agent's code block, test the draft, then pin it as a new version.”

1.13.0 - Freeze keeps repeatable steps as pinned code by itself

2026-10-12

  • agentsFreeze now decides what becomes code: loops, data shaping and tool calls with clear rules turn into Python code blocks the frozen agent runs, while steps that need judgment stay with the agent. You just freeze; the saved instruction lists each code block with its version and SHA-256.
    Use: freeze_agent, get_compilation, run_frozen - docs
    Try: “Freeze my Hacker News topic counter and show me which steps became code.”
  • agentsFrozen code does not drift: editing a code block saves a new version, and every frozen version keeps running the exact code it was frozen with. The code block view in the app shows each version and which version every run used.
    Use: run_frozen, save_instruction_version, set_live_version
    Try: “Edit my agent's code block, then run its live frozen version and show which code version ran.”

1.12.0 - Keep every version of a frozen agent and choose the live one

2026-10-11

  • agentsChange a frozen agent without losing what works: save_instruction_version saves your edited instruction as a new version and keeps the old one, run_frozen tests it, and set_live_version makes it live. Runs and schedules that name no version use the live one; set an older one live to roll back.
    Use: save_instruction_version, set_live_version, run_frozen, list_compilations
    Try: “Save my site tester's edited instruction as a new version and keep its current live version.”
  • sdkPython SDK 1.2.0rc13 (pip install 'flymyai>=1.2.0rc13'): client.compilations.save_instruction, set_live, get_live and run_live. Over REST the same routes, and is_live=true on the compilation list finds the live version.
    Use: /api/v1/agents/compilations/save-instruction/{execution_id}/, /api/v1/agents/compilations/{id}/set-live/, /api/v1/agents/compilations/live/{agent_id}/run-instruction/ - docs
    Try: “Show my site tester's live saved version using the FlyMyAI Python SDK.”
  • agentsupdate_compilation pauses and resumes a version's schedule with schedule_paused, keeping its cron, labels a version with name and note, and narrows or revokes its consent for browser screenshots and files.
    Use: update_compilation, get_compilation
    Try: “Label a saved version of my site tester forms check and show its browser consent.”
  • mcpupload_agent_file attaches a small file - a script, a prompt, a CSV - to an agent in one call, and the same operation_key returns the same file. The agent reads it in its sandbox or by name in its prompt.
    Use: upload_agent_file, list_agent_files
    Try: “Attach a small JSON file to my site tester agent.”

1.11.3 - Code steps read recent tool results; AI Overview sources; saved browser consent

2026-10-10

  • agentsA code step can mount its recent tool results as files: run_tool_responses=true on execute_code or execute_code_with_outputs supplies the saved results within the sandbox limits, and a manifest names the tool behind each file.
    Use: execute_code, execute_code_with_outputs - docs
  • toolsGoogle results from DataForSEO can include the AI Overview with the pages it cites: set load_async_ai_overview on dataforseo_serp_google_organic_live.
    Use: dataforseo_serp_google_organic_live - docs
  • billingDataForSEO on FlyMyAI's account costs exactly what DataForSEO charges for each call, 0% markup. With your own DataForSEO key you pay only compute time.
  • agentsA frozen agent keeps your consent for the browser screenshots and files its source run saved, so its runs and its schedule finish without asking. The app shows that consent on the version, and Ask every run takes it back.
    Use: freeze_agent, run_frozen, get_compilation
    Try: “Show the browser consent saved on a frozen version of my site tester.”
  • apiUpload a file to an agent with one JSON request - filename with content or content_base64, up to 1 MiB - on the route that takes multipart uploads; the same Idempotency-Key returns the same file.
    Use: /api/v1/agents/tasks/{uuid}/files/upload/

1.11.2 - Seedream 5.0 Pro has an API safety-checker option

2026-10-09

  • modelsSeedream 5.0 Pro creates or edits an image from a prompt and references. Its API accepts enable_safety_checker, enabled by default; explicit false selects the alternate image route. Check get_pricing for the price of each setting.
    Use: get_model_schema, get_pricing, run_model - docs
    Try: “Use Seedream 5.0 Pro to make this portrait's background blue with enable_safety_checker=false.”

1.11.1 - Choose Grok 4.7 as your agent brain

2026-10-09

  • agentsChoose Grok 4.7 as an agent brain in the model picker or from the Agents MCP. Set its reasoning level, let it call tools and keep working after context compression. Each brain call uses your FlyMy balance.
    Use: list_agent_models, create_agent, run_agent - docs
    Try: “Create an agent on Grok 4.7 that returns a JSON object with status set to ok, then run it.”
  • apiGrok 4.7 supports function tools and tool-result messages through its OpenAI-compatible API. Send a stable prompt_cache_key for a continuing chat.
    Use: get_model_schema - docs
    Try: “Show me the Grok 4.7 OpenAI-compatible base URL.”

1.11.0 - Change an artifact by pasting one line into Claude; start one from a standard type

2026-10-09

  • artifactsChange an artifact by pasting one line into Claude: copy "<link> add these things to the artifact: " from its page and type the change. If you may edit it, you get a new version next to the live one and both stay; end with "and update current version" to make it live. Others get their own copy.
    Use: get_artifact, read_artifact_files, publish_artifact_version, clone_artifact - docs
    Try: “Make v2 of my Golden Harvest game with different enemies, and keep v1 live.”
  • artifactsStart an artifact from a standard type: a presentation, a 3D game, a landing page, a report, a dashboard or a gallery. Each works as it is and keeps its content in one file that its README.md explains, so most changes touch only that file.
    Use: list_artifact_types, create_artifact, /api/v1/artifacts/types/ - docs
    Try: “Make me a FlyMyAI presentation about our Q3 results, with a slide on pricing.”
  • artifactsSave a new version next to the live one without replacing it, open it by its page link (?v= and the version number) and make it live when you are ready. get_artifact takes the artifact's own page link as well as a share link, and says what you may do next.
    Use: publish_artifact_version, update_artifact, get_artifact - docs
    Try: “Save a dark version of my landing page next to the live one and send me its link.”
  • artifactsFiles that look like secrets stay out of your pages: env files, version control folders, keys and certificates, and text files holding another service's token are refused with the file's name, because everyone who opens a page can read its files.
    Use: create_artifact, publish_artifact_version - docs
  • sdkPython SDK 1.2.0rc12 (pip install 'flymyai>=1.2.0rc12'): client.artifacts.types() lists the standard types, create(type=...) starts from one, and a share link with a query string or the artifact's own page link works wherever the SDK takes a link or an id.
    Use: /api/v1/artifacts/types/, /api/v1/artifacts/ - docs
    Try: “Write a Python script with the FlyMyAI SDK that lists the artifact types and creates a dashboard.”

1.10.14 - Build and run helper agents from your app chat

2026-10-08

  • agentsIn your FlyMyAI app chat, your agent can create or edit your agents using your connected tools, then run them as subagents. New chats use a shared $5 subagent budget by default; you can change it before running.

1.10.13 - Seedance 2.5: video with sound up to 30 s, real people in frames and references

2026-10-08

  • modelsSeedance 2.5 by ByteDance makes videos with sound of 4 to 30 seconds at up to 1080p from a prompt, a first and last frame, or up to 30 reference images, 10 clips and 10 audio tracks; frames and references may show real people. A 5-second 720p clip costs about $1.22.
    Use: recommend_model, get_pricing, run_model, /api/v1/flymyai/bytedance-seedance-2_5-reference-to-video/predict/async/ - docs
    Try: “Make a 5-second 720p video with Seedance 2.5 of a paper boat drifting along a rain-soaked street at dusk.”

1.10.12 - YoTrends: what is trending on YouTube and TikTok

2026-10-08

  • toolsYoTrends finds what is trending on YouTube and TikTok: viral videos by platform and region, the creators that trend most, and content categories with their trend weight and trendlines. It runs on FlyMyAI's YoTrends plan with nothing to set up; a call costs compute time.
    Use: search_tools, execute_tool - docs
    Try: “What is trending on TikTok in the US this week, and which creators are behind it?”

1.10.11 - Gemini and Grok text models to pick and run; Instagram albums list every slide

2026-10-08

  • modelsGemini 3.1 Flash Lite, Gemini 2.5 Flash, Gemini 3 Flash, Gemini 3.1 Pro and Grok 4.1 Fast are in the model catalog as chat models, priced per 1M tokens: recommend_model suggests them, run_model runs them, and agents in the app can be given them in the model picker.
    Use: recommend_model, flymyai_models, get_pricing, run_model
    Try: “Find the cheapest text model on FlyMyAI and have it write a limerick about a cat who codes.”
  • modelsrun_model answers with text models that only stream, such as Gemini and Grok 4.1: one paid call returns the whole answer and its token usage.
    Use: run_model
    Try: “Run flymyai/google-gemini-31-flash-lite-preview: three names for a coffee shop in Lisbon.”
  • toolsinstagram_post_info lists every slide of an album post in order, with each image or video URL, ready to pass to analyze_media.
    Use: instagram_post_info, analyze_media - docs
    Try: “Describe every slide of the latest album post on @natgeo.”

1.10.10 - Nano Banana 2.1: Google's newest image model, up to 4K

2026-10-07

  • modelsNano Banana 2.1 by Google makes and edits images up to 4K from a prompt and up to 14 reference images, with readable text and characters that keep their look across edits. You pay Google's token price: $0.0336 per 1K image plus thinking tokens, about $0.04 a run.
    Use: recommend_model, get_pricing, run_model, /api/v1/flymyai/nano-banana-2_1/predict - docs
    Try: “Make a 2K poster for a jazz night in Lisbon titled "Noite de Jazz" with Nano Banana 2.1.”

1.10.9 - Leads use their subagents' files in code; Instagram in Agent Teams at the provider's price

2026-10-07

  • agentsA lead agent sees and reads the files its subagents save and can pass them straight into its code run: the code finds them under /tmp/workspace/inputs, so collected data is processed without copying it by hand.
    Use: run_agent, get_run - docs
    Try: “Have a subagent save a CSV of sales by region, then total it in code and tell me the sum.”
  • toolsInstagram data in Agent Teams on FlyMyAI's shared account costs the same as in a single agent: $0.02 per request, what the data provider charges. With your own HikerAPI key you pay only compute time.
    Use: run_agent - docs

1.10.8 - Subagents hand back what they have, and leads finish with partial results

2026-10-07

  • agentsA subagent that reaches its budget or keeps repeating the same call is stopped with its saved result and the reason, and its lead finishes with what the subagents delivered instead of redoing their work or waiting.
    Use: run_agent, get_run, get_team_run - docs
    Try: “Have three subagents each summarize one of these three articles, and give me what they finish.”
  • agentsAn API, frozen or scheduled run that cannot do part of its task, such as reading a file that does not exist, finishes as completed and names the missing part instead of failing.
    Use: run_agent, run_frozen, schedule_agent - docs

1.10.7 - Pick the model your project's agent runs on

2026-10-07

  • appChoose the model of a project's agent: the Model list in the project panel, `model` in projects_agent, the REST endpoint or the Python SDK 1.2.0rc11. A new project agent starts on the strongest model, and your choice stays through platform updates.
    Use: projects_agent, list_agent_models, /api/v1/agents/projects/{id}/agent/ - docs
    Try: “Switch my project's agent to GPT-5.6 Terra.”

1.10.6 - Instagram at the provider's price, project spend with its agent, vision refusals say why

2026-10-07

  • toolsInstagram data on FlyMyAI's shared account costs what the data provider charges for each request: $0.02 per request, and the balance check is free. With your own HikerAPI key you pay only compute time.
    Use: search_tools, execute_tool - docs
    Try: “Get the follower counts of @nasa and @spacex.”
  • appA project's spend now includes its project agent: the model and tool calls of its chats and of the subagents it starts count next to page calls and servers.
    Use: projects_get, projects_list - docs
    Try: “How much did my project spend this week, including its agent's chats?”
  • modelsWhen a vision model refuses a request, the answer says why in a reason field next to the usual error, for example a prompt longer than 100,000 characters.
    Use: analyze_media, /api/v1/flymyai/gemini-3_8-flash-vlm/predict - docs

1.10.5 - Share links open straight to the page, and your own version is one prompt away

2026-10-06

  • artifactsA shared artifact link now opens the page under one thin bar. With sources shared, Make your own version gives copy-ready steps for Claude Code, Codex and chat apps: connect FlyMy.AI, paste one prompt, and your agent clones it and publishes your version. Agents fetching the link see them too.
    Use: clone_artifact, read_artifact_files, publish_artifact_version, /api/v1/artifacts/shared/{handle}/ - docs
    Try: “Make my own version of the FlyMy.AI artifact my colleague shared with me and change its title.”

1.10.4 - Python SDK: add your own MCP servers with OAuth sign-in

2026-10-06

  • sdkPython SDK 1.2.0rc9 (pip install 'flymyai==1.2.0rc9') adds client.mcp_servers: add an MCP server by URL, sign in to an OAuth server with a link a person approves, then call its tools from your code. Tokens and the client secret never come back.
    Use: /api/v1/agents/mcp-servers/, /api/v1/agents/mcp-servers/{id}/authorize/ - docs
    Try: “Write a Python script with the FlyMyAI SDK that adds Linear's MCP server with OAuth and prints the sign-in link.”

1.10.3 - Slack as you: search, DMs, files, canvases and lists

2026-10-06

  • toolsConnect Slack once and agents work as you: search messages and files, read and post in channels and direct messages, share files, write canvases and lists, and set reminders. The bot connector is now called Slack Bot and posts as the FlyMyAI app.
    Use: search_tools, execute_tool - docs
    Try: “Search my Slack for messages about the release this week and summarize what was decided.”

1.10.2 - Wan 3.0 video with sound, up to 30 seconds

2026-10-06

  • modelsWan 3.0 by Alibaba makes video with sound from a prompt, from a first and last frame, or from reference images, clips and audio: 2-30 seconds or a length it picks, at 480P, 720P or 1080P. You pay per generated second, $0.05 to $0.20 by resolution.
    Use: recommend_model, get_pricing, run_model, /api/v1/flymyai/wan-3_0-t2v/predict/async/ - docs
    Try: “Make a 10-second 720P video with sound of a red fox running through fresh snow at dawn with Wan 3.0.”

1.10.1 - Grok 4.7 with private mode

2026-10-06

  • modelsGrok 4.7 by xAI answers prompts and chats with text and images through /predict, the OpenAI-compatible API and run_model. Choose how hard it reasons, from low to xhigh, and turn on private mode to keep the prompt and the answer out of FlyMyAI storage.
    Use: run_model, get_pricing, /api/v1/flymyai/grok-4_7/openai/v1/chat/completions, /api/v1/flymyai/grok-4_7/predict - docs
    Try: “Ask Grok 4.7 in private mode to rewrite my launch email in three tones.”

1.10.0 - Sign in to OAuth MCP servers, and a Collaborator.pro connector

2026-10-05

  • mcpAdd an MCP server that signs in with OAuth, such as Linear's or Notion's: FlyMyAI registers itself, you approve access in the provider's window, and access renews on its own. A server without automatic registration takes the client ID and secret of your own OAuth app.
    Use: add_mcp_server, authorize_mcp_server, connect_mcp_server - docs
    Try: “Connect Linear's MCP server https://mcp.linear.app/mcp with OAuth and list my open issues.”
  • toolsConnect Collaborator.pro with your API token. Agents read your link-building deals as an advertiser or a site owner, and search its catalog of sites for guest posts by price and SEO metrics once Collaborator opens the catalog for your account.
    Use: search_tools, execute_tool - docs
    Try: “Show my Collaborator deals with their status, site and price.”

1.9.2 - Instagram posts with shares, reposts and saves

2026-10-05

  • toolsInstagram Public Data returns an account's photos, albums and reels in one list: pinned posts first, then newest first. Posts and reels include shares; posts also include reposts and saves when Instagram returns them.
    Use: search_tools, execute_tool - docs
    Try: “Show the latest posts of @nasa with likes, comments and shares, sorted by shares.”

1.9.1 - Clone your app projects from an assistant

2026-10-05

  • mcpAsk your assistant to clone an app you own. Preview its latest successful release, then create a private copy with new resources and a separate budget. Stored data, chats and personal connections stay with the original.
    Use: projects_plan, projects_create - docs
    Try: “Clone my app project as a private copy with a separate budget after I approve its plan.”

1.9.0 - Navigate projects, clone apps and recover agent context

2026-10-05

  • appProject workspaces use the full page width. Move between sections from the Projects menu, and drag the divider to adjust chat width beside the graph. Your browser remembers the width; the account menu stays on the right.
    Use: projects_list - docs
    Try: “Show me my projects so I can open one in the workspace.”
  • appClone your own app in its workspace. Review the plan, then create a private copy with new resources and its own budget. The copy uses the latest successful release; stored data, conversations and personal connections are not copied.
    Use: projects_list, projects_get - docs
    Try: “Find my app project and open its workspace so I can make a private copy.”
  • agentsCloud agents save private checkpoints before compacting a conversation. Recovery pointers let them retrieve earlier decisions, evidence and unfinished work in bounded chunks; ordinary chats can find prior checkpoints of the same agent.

1.8.0 - Project workspace with a master chat and agent history

2026-10-04

  • appOpen a project with a searchable project picker, one agent list, a resource graph and a visible master chat. Select an agent to read its conversation and earlier runs without leaving the workspace.
    Use: projects_list, projects_get - docs
    Try: “Show my projects and give me a link to the project workspace.”

1.7.0 - Projects from one config, and web pages you can version and share

2026-10-03

  • apiA project is one flymy.yaml: its pages, agents, storage and budgets. Plan it first to see what it creates and its price per hour; nothing is created until you confirm, and every change you apply is a new release.
    Use: projects_plan, projects_create, apps_plan, apps_apply, projects_list - docs
    Try: “Make me a FlyMyAI project called notes with a $5 daily budget and show me the plan first.”
  • artifactsPublish a small web page, game or deck as an artifact. Every change is a new version and you pick the live one; share it by link, or invite people by email or username to view or edit it.
    Use: create_artifact, publish_artifact_version, share_artifact, add_artifact_member, clone_artifact - docs
    Try: “Turn this HTML into a FlyMyAI artifact and invite alex@example.com to edit it.”
  • sharingAn artifact someone shares with you by name waits as an invitation: nothing of it opens until you accept, and you can decline. The owner sees you only as the email or username they typed.
    Use: list_artifact_invitations, accept_artifact_invitation, decline_artifact_invitation - docs
    Try: “Show the artifact invitations waiting for me.”
  • appPublish a page as a site and share it by link or with people by email. Its declared calls to models and agents are paid by your sponsor pool or by visitors who agree to pay, within daily caps.
    Use: publish_site, share_site, invite_to_site, get_site_usage - docs
  • billingDirect tool calls - execute_tool, the REST tool call, and compiled and scheduled runs - are charged like agent runs. A paid call needs a positive balance; with an empty balance it is refused before it runs.

1.6.6 - See why a scheduled run failed and act on it from its page

2026-10-03

  • appA failed scheduled run's page says why it failed and lets you change the instruction of the version that ran (the schedule stays on it), run it again or resume the schedule.
    Try: “Open my last failed scheduled run and tell me what to change in its instruction.”
  • agentsFailure emails open the failed run itself, quote the agent's last message and say how many runs before it succeeded. A run that stops without a result reports the agent's own last words as its error.
    Use: get_run, read_run_result
    Try: “Why did my daily digest fail this morning?”

1.6.5 - Scheduled agents tell you when they fail

2026-10-02

  • agentsWhen a scheduled run fails, the agent's owner gets an email with the reason and a link to act on it: reconnect a connection, top up the balance or open the agent.
    Try: “Check my scheduled agents and tell me if any of them failed this week.”
  • agentsAfter 3 failed scheduled runs in a row the schedule pauses itself and keeps its time. get_compilation shows schedule_paused and the reason; resume from the agent page, with schedule_paused false, or by setting the cron again.
    Use: get_compilation, update_compilation, PATCH /api/v1/agents/compilations/{id}/ - docs
    Try: “I reconnected Telegram - resume my paused daily news agent.”

1.6.4 - Share from Claude, ChatGPT and Grok

2026-10-02

  • sharingEvery sharing action works from the FlyMyAI connection in Claude, ChatGPT or Grok: see what you can share, share by email or with a teammate, accept, decline and revoke. list_shareable_mcp_connections and list_mcp_share_people help pick what and with whom.
    Use: list_shareable_mcp_connections, list_mcp_share_people, invite_mcp_user, revoke_mcp_access - docs
    Try: “Share my Instagram with alex@example.com for 7 days.”
  • sharingInstagram, Facebook, LinkedIn, Reddit, TikTok and YouTube connections can be shared like any other connection.
    Use: invite_mcp_user
    Try: “Share my LinkedIn with my teammate so they can draft posts for me.”
  • sharingSee every call made through a share - what ran, when and what it cost: Calls in Shared by me, Your calls in Use, or list_mcp_share_activity in chat.
    Use: list_mcp_share_activity - docs
    Try: “Show the last calls Alex made through the Notion I shared.”

1.6.3 - Shared connections and fleets in your assistant

2026-10-02

  • sharingConnections other people share with you work in Claude, ChatGPT, Grok and any MCP client signed in to FlyMyAI: list_configured_tools lists them next to your own under shared_with_me, with who shared each, and call_shared_mcp calls them. No separate token or snippet.
    Use: list_configured_tools, list_shared_mcp_tools, call_shared_mcp - docs
    Try: “Read the Notion page my teammate shared with me and give me three bullet points.”
  • sharingA connection shared with you has no monthly call cap: you pay for your own calls from your balance until the owner revokes it or the access period ends.
    Use: call_shared_mcp
    Try: “Check the open Linear issues in the workspace Alex shared with me.”
  • agentsYour fleets are on the Agents MCP: projects_list shows each fleet - a lead agent and the agents it starts - with its status, runs, spend and failed runs today, this week and this month; projects_get opens one with its agents, schedules, tools and diagram; projects_errors reads its errors.
    Use: projects_list, projects_get, projects_errors
    Try: “Show my fleets and what each one spent this week.”

1.6.2 - Connector calls at $0.000037 per second

2026-10-02

  • billingTool calls priced by running time - Notion, Slack, GitHub, Google, CRMs and other connectors, including calls through a connection shared with you - cost $0.000037 per second ($0.133 per hour). Search, media and voice tools keep their own price.
    Use: execute_tool, call_shared_mcp - docs
    Try: “Read my latest Notion page and turn it into three bullet points for Slack.”

1.6.1 - Sandbox and media editing at $0.0666 per hour

2026-10-01

  • billingCode in the sandbox and audio or video edits with media_editor cost $0.0000185 per second ($0.0666 per hour). execute_code and media edits are billed for the seconds the call takes, run_code_block only for its execution segment.
    Use: execute_code, execute_code_with_outputs, run_code_block, create_video_from_images, replace_audio - docs
    Try: “Run Python that simulates 100,000 dice rolls and chart how often each total comes up.”
  • modelsGPT-5.6 Sol costs $4 input and $20 output per 1M tokens. On GPT-5.6 models, input written to the prompt cache bills at 1.25x the input rate, as on GPT-6 and Claude.
    Use: list_agent_models - docs
    Try: “Compare GPT-5.6 Sol and Claude Sonnet 5 for a long research agent and pick the cheaper one.”

1.6.0 - Your agent fleets are projects: a diagram, money and errors for each lead and its agents

2026-10-01

  • appProjects lists each fleet - a lead agent and the agents it starts as subagents or in its Team - with a diagram of the lead, its agents, their models, tools, schedules and budget, and what they ran and spent today, this week and this month.
    Use: GET /api/v1/agents/projects/, GET /api/v1/agents/projects/fleet:<lead id>/ - docs
    Try: “Open Projects and show me what my marketing fleet spent this week, agent by agent.”
  • agentsA fleet's errors journal groups the failed runs of each of its agents with the newest error and a link to that run, and the Agents tab says which fleet a chat belongs to.
    Use: GET /api/v1/agents/projects/fleet:<lead id>/errors/
    Try: “Which agent in my fleet failed most this week, and why?”

1.5.2 - Browser tasks default to GPT-6 Luna

2026-10-01

  • toolsNew browser sessions use gpt-6-luna by default. Pick a newer browser model with model, for example gpt-6-sol or claude-opus-5. code_mode and use_own_llm_key keep working on gpt-5.6-luna.
    Use: run_browser_task, send_task - docs
    Try: “Fill out the practice form at demoqa.com/automation-practice-form for Ada Lovelace and show me the confirmation.”

1.5.1 - Share a connection with no end date from chat

2026-10-01

  • sharinginvite_mcp_user shares a connection with no end date: access_days 0 keeps access until you revoke it. 1 to 30 days work as before.
    Use: invite_mcp_user, revoke_mcp_access
    Try: “Share my Notion with alex@example.com with no end date.”
  • sharingcancel_mcp_invitation also declines an invitation sent to you; its link then stops offering Accept.
    Use: cancel_mcp_invitation
    Try: “Decline the Calendly invitation I got today.”

1.5.0 - Share any connected service with another person, reads and writes

2026-09-29

  • sharinginvite_mcp_user shares a connection of any supported service by email. The recipient can use every one of its actions, reads and writes, on your account; they pay for their own calls, your credentials stay with FlyMyAI, and you can revoke it at any time.
    Use: invite_mcp_user, list_shared_mcp_tools, call_shared_mcp
    Try: “Share my Linear with alex@example.com for 7 days so they can create and update issues.”
  • sharingA recipient's own agents can use the actions of connections shared with them, reads and writes, in their runs.
    Use: run_agent
    Try: “Make an agent that files a Linear issue in the workspace my teammate shared with me.”
  • sharingTeam connections share every action the owner selects, reads and writes. Built-in FlyMyAI tools and tools that run on FlyMyAI's own key are not shared.
    Try: “Share my Gmail with my team so our agents can draft replies.”

1.4.1 - Gemini 3.8 Flash vision reads images at the resolution you choose

2026-09-29

  • modelsgemini-3_8-flash-vlm takes media_resolution low, medium or high. low reads an image with about a quarter of the input tokens; a video is already read at low by default.
    Use: analyze_media, /api/v1/flymyai/gemini-3_8-flash-vlm/predict
    Try: “Describe these 20 product photos at media_resolution low and give me one line each.”

1.4.0 - Share an agent team: a clone of a lead copies the agents it runs

2026-09-29

  • sharingA public lead agent is cloned together with the agents it runs as subagents, all or none, and the copy starts your copies. The share page lists each agent with its model and tools.
    Use: copy_agent, set_agent_public
    Try: “Make my lead agent public so a friend can clone it with its three agents.”
  • modelsgemini-3_8-flash-vlm takes a prompt and a system_instruction of up to 100,000 characters each.
    Use: analyze_media, /api/v1/flymyai/gemini-3_8-flash-vlm/predict
    Try: “Label these three ad videos with my labeling guide and JSON schema.”
  • billingwhoami shows pending_usd, the usage still being charged, and available_usd, the balance after it.
    Use: whoami
    Try: “How much of my balance is still pending?”
  • billingA run's price shows completion review, tool discovery and context compression on their own lines, and lists the models the run ran through run_model.
    Use: get_execution_price
    Try: “What did my last run spend on its own model and on service calls?”

1.3.2 - Open generated media from chat, and tell freeze what to keep as is

2026-09-29

  • mcprun_model gives each generated file a url that opens it without signing in for up to 15 minutes, so an image made in chat can be viewed and saved right away.
    Use: run_model
    Try: “Make a flat blue circle icon with nano-banana and give me a link to open it.”
  • mcpanalyze_media takes the Gemini 3.8 Flash controls: response_schema, thinking_level, temperature and seed, and fps, start_offset and end_offset for a video.
    Use: analyze_media
    Try: “Describe the first 5 seconds of this video as JSON with hook and on_screen_text: https://example.com/ad.mp4”
  • agentsfreeze_agent and schedule_agent take guidance: plain text the compiler follows, such as which steps and inputs must stay as they are.
    Use: freeze_agent, schedule_agent
    Try: “Freeze my scout agent and keep the budget as an input and its three search steps as they are.”

1.3.1 - Gemini 3.8 Flash vision follows your JSON Schema and lets you choose how much it thinks

2026-09-29

  • modelsgemini-3_8-flash-vlm takes response_schema, a JSON Schema its answer must follow, and thinking_level low, medium or high. low spends fewer thinking tokens, which suits labeling and extraction.
    Use: /api/v1/flymyai/gemini-3_8-flash-vlm/predict
    Try: “Label this ad video with my JSON schema at thinking_level low: https://example.com/ad.mp4”
  • modelsIts answers include finish_reason, so MAX_TOKENS shows that max_output_tokens cut the answer, and usage.thinking_tokens. temperature and seed tune sampling; for a video, fps, start_offset and end_offset choose what is read.
    Try: “Read only the first 3 seconds of this video at 1 frame per second and tell me the hook.”
  • apiThe Models API names every field this model does not use in warnings instead of ignoring it.
    Use: /api/v1/flymyai/gemini-3_8-flash-vlm/predict

1.3.0 - A lead agent runs your other agents as subagents, each with its own model and tools

2026-09-29

  • agentsA run can start your other agents as subagents with delegate_agent: each works with its own model, prompt, tools and MCP servers, and the lead collects their results. Choose which agents it may start in subagent_limits.agents or in the chat Subagents control.
    Use: run_agent, run_frozen
    Try: “Make a lead agent that asks my research agent and my copywriter agent for three ad ideas, then picks the best one.”
  • agentsdelegate_agent without an agent runs a copy of the lead with its tools on another model from list_agent_models, for example a cheaper model for simple parallel steps.
    Use: list_agent_models
    Try: “Split this research into three parts and run each part in parallel on gemini-3.8-flash.”
  • agentsEach subagent gets its share of the run's subagent budget and stops when it is spent. get_team_run shows every subagent's agent, model and spend, and subagent_status returns a large result in full.
    Use: get_team_run
    Try: “Run my lead agent with a $2 subagent budget and show me what each subagent spent.”
  • agentsA lead can freeze and schedule its subagents with freeze_subagent and schedule_subagent. A frozen or scheduled lead starts the frozen versions of its agents that it pinned when it was frozen, so editing an agent does not change a schedule.
    Use: freeze_agent, schedule_agent, get_compilation
    Try: “Freeze my scout agent from its last run and schedule it every weekday at 9:00 in my timezone.”

1.2.0 - See what a model run costs before you run it, and call chat models with the OpenAI SDK

2026-09-28

  • modelsModel pages and model cards show each model's price, computed from the same rules that charge the run: per run, per second of video, per image, per 1M tokens or per hour of run time, with a table when inputs like resolution or duration change it.
    Try: “How much does an 8-second Veo 3.1 video cost at 1080p with sound?”
  • mcpget_pricing quotes any listed model, and get_model_schema, flymyai_models and recommend_model show the same price: one run with default settings, the range and the inputs that change it.
    Use: get_pricing, get_model_schema, flymyai_models, recommend_model
    Try: “Find the cheapest way to make a 5-second 720p video and tell me what it will cost.”
  • apiModel endpoints in the REST API return price, and chat models also return openai.base_url for the official OpenAI SDK. GET /api/v1/demo/catalog/ lists every public model with both.
    Use: /api/v1/demo/catalog/ - docs
    Try: “List the public models that cost under $0.05 per run.”
  • appChat model pages have an OpenAI tab with ready Python, JavaScript and cURL code for the OpenAI-compatible route: keep your OpenAI code and change the base_url and key.
    Try: “Show me how to call Gemini 2.5 Flash with the OpenAI Python SDK.”

1.1.1 - Read a run's whole result and see why a direct tool call was refused

2026-09-28

  • agentsread_run_result reads the whole result or error of a finished run when it is longer than the preview get_run shows, in parts of up to 64 KiB, using the ref from get_run. The sha256 from get_run confirms you have every byte.
    Use: get_run, read_run_result - docs
    Try: “Run my research agent and give me its whole report, not just the first part.”
  • toolsWhen a direct execute_tool call is turned down before anything happens - wrong arguments or filters, missing setup, or the service refusing the request - the answer says rejected and gives the reason. Correct the call and send it again with a new operation_key.
    Use: execute_tool - docs
    Try: “Create a Linear issue for this request and, if Linear turns it down, tell me exactly why.”
  • agentsrun_agent and run_frozen answer right away with a short receipt: the run id, status, run_seq and subagent limits. Follow the run and read its output with get_run.
    Use: run_agent, run_frozen, get_run - docs
    Try: “Start my headline agent on these ten product links and tell me when it is done.”
  • toolslist_configured_tools returns one short row per connection - id, alias, toolkit and status - 25 to a page by default. Filter by toolkit or pass view=full to get setup steps and the connect link.
    Use: list_configured_tools - docs
    Try: “List my connected tools and tell me which ones still need setup.”

1.1.0 - Shared MCP connections in My MCPs and in your agents

2026-09-28 - pilot

  • sharingConnections other people share with you appear in My MCPs next to your own, with who shared them and their email. Accept or decline an invitation right there; the owner gets an email when you accept. Pilot: Linear, Notion, Calendly and Gmail reads.
    Use: list_mcp_shares, list_mcp_invitations, accept_mcp_invitation - docs
    Try: “Show the MCP connections people shared with me and who shared each one.”
  • sharingShare a connection for 1, 7, 14 or 30 days, or in the app with no end date until you revoke it.
    Use: invite_mcp_user - docs
    Try: “Share my Notion connection with alex@example.com for 14 days.”
  • agentsAgents use your own connections first and can use connections shared with you: with no Notion of your own, a Notion request uses the shared one; ask for someone else's or a shared connection to get theirs. Calls are billed to your balance.
    Try: “Find the latest pages in the Notion that alex@example.com shared with me.”

1.0.1 - Baseline: FlyMyAI Headless Cloud as it runs in production

2026-09-25

  • modelsOne catalog of generative models: images, video, audio and music, speech, upscaling, embeddings and LLMs. recommend_model picks one for a task, get_pricing shows the price, run_model runs it on your key.
    Use: list_media_models, recommend_model, get_pricing, run_model - docs
    Try: “Make a 5-second vertical video of neon koi fish swimming through rain.”
  • agentsBuild one agent from an idea: create it, run it and read the run, refine the same agent, then freeze an accepted run into a reusable release and run that release. A frozen agent keeps running in the cloud when your laptop is closed.
    Use: create_agent, run_agent, get_run, freeze_agent, get_compilation, run_frozen - docs
    Try: “Build me an agent that turns a product link into three ad headlines, test it and freeze it.”
  • templatesOpen a media template from the gallery at flymy.ai/media (product ads, unboxings, outfit reveals and more), run it as is, clone it and make it your own agent.
    Try: “Show me the media templates and clone the unboxing one for my sneakers.”
  • toolsSearch the connector catalog by what you need done (Slack, Notion, GitHub, Google Workspace, CRMs, payments, web search, browser automation and more), read the exact schema, then run an action directly or inside an agent.
    Use: search_tools, get_tool_schemas, execute_tool, add_tool - docs
    Try: “Post a summary of my open Linear issues to our team Slack channel.”
  • sharingPilot: share one connected account with another person by email from the MCP page in the app. They use only reviewed read actions and pay for their own calls; your credentials never leave FlyMyAI. Sharing with a team is in a private pilot.
    Try: “Share my Notion connection with alex@example.com for a week.”
  • templatesShare an agent as a template: set_agent_public makes it public, and copy_agent forks an agent into your own account to change as you like.
    Use: set_agent_public, copy_agent
    Try: “Make my headline agent public so a friend can fork it.”
  • agentsRun a frozen agent on a schedule, or call it from your product over the Agents REST API or the Python SDK (package flymyai).
    Use: schedule_agent, run_frozen - docs
    Try: “Run my headline agent every morning at 9 and post the result to Slack.”
  • agentsAn agent can hand parts of a run to subagents and join their results in the same run. subagent_limits on run_agent caps their spend, count and parallelism before anything starts.
    Use: run_agent
    Try: “Research these five competitors in parallel and give me one comparison table.”
  • modelsTypeSafe Jev 1.13 answers typed questions quickly and cheaply: pick one option, give a yes probability or a score. Use it for routing, triage and guardrail checks from an agent or directly.
    Use: make_typed_decision - docs
    Try: “Sort these support tickets into billing, technical and sales.”
  • toolsConnect several accounts of one service under aliases, group connections into MCP resource sets and grant a set to agents or to an agent group.
    Use: add_tool, create_mcp_resource_set, replace_mcp_resource_set_members, create_agent_group - docs
    Try: “Connect my work and personal Gmail and let my inbox agent use only the work one.”
  • mcpConnect an MCP client to one server, https://mcp-agents.flymy.ai/mcp, with OAuth sign-in or an API key sent as X-API-Key. It carries models, tools and agents; whoami confirms which account is connected.
    Use: whoami, search_tools - docs
    Try: “Which FlyMyAI account am I on, and what is my balance?”
  • mcpBring your own MCP server: register it, connect it and call its tools from chat or from an agent.
    Use: add_mcp_server, connect_mcp_server, call_mcp_server - docs
    Try: “Connect my MCP server at https://example.com/mcp and list its tools.”
  • apiEmbed an agent in your product with one builder key: publish a frozen agent as a deployment, identify each of your users by external_user_id, let them connect their own accounts through a hosted link, and run and meter per user.
    Use: POST /api/v1/agents/deployments/{deployment_id}/run/ - docs
    Try: “Show me the code to call my frozen agent from my Next.js app.”
  • computeDeploy your own Python model as an auto-scaling GPU endpoint with the FlyMyAI CLI (fma) and pay only for the compute it uses.
    Try: “How do I deploy my PyTorch model as a GPU endpoint?”