Skip to content
Back

AI posts

AI release notes

Official updates from Anthropic, OpenAI, Google, xAI, Mistral, DeepSeek and Meta, newest first. Open any update for a short AI summary and the full notes.

Compiled by the Toolstash team · Updated

  1. xAI · New modelAnnouncement

    Grok 4.7

    Grok 4.7, SpaceXSI's frontier model for coding, agentic tasks, and knowledge work, is now available on the xAI API as grok-4.7. It has a 500k context window, text and image inputs with text-only output, and no text output limit. Pricing is $2 / $0.50 / $6 per 1M tokens (input / cached input / output) below 200k prompt tokens, and $4 / $1 / $12 above.

  2. xAI · New model

    Grok Voice Transcribe 2.0

    grok-voice-transcribe-2.0 is now available. Use grok-voice-transcribe-1.0 or grok-voice-transcribe-2.0; the default is grok-voice-transcribe-2.0.

  3. Anthropic · Claude Code

    Claude Code 2.1.252

    4 fixes

    Fixed Bash commands failing with "task output swap refused (tasks dir moved or linked)" on some Macs. Fixed "always allow" not saving in a project that has no.claude/settings.local.json yet.

  4. OpenAI · Codex app

    Codex app

    Performance improvements and bug fixes. Fixed new chats getting stuck after pasting a large amount of text. Improved startup recovery on Windows and cleared stale startup errors after reconnection. Fixed response streaming stalls in hidden app windows.

  5. Google · Antigravity

    Default Gemini 3.8 Flash model upgrade, lightweight agent configuration, and Vertex AI Express mode

    The 0.1.16 release updates the default model for new agents to gemini-3.8-flash for enhanced reasoning capabilities, introduces a.lightweight() method on agent configs for small and local models, adds Vertex AI Express mode with API keys, and expands tool runner support for callable classes and dataclass functors.

  6. Mistral AI · New model

    OCR 4.1 is now Generally Available

    OCR 4.1 (mistral-ocr-4-1) is now Generally Available.

  7. OpenAI · Codex CLI

    Codex CLI 0.151.0

    3 new6 fixes2 other

    Added a configurable grace period for discovering tools from optional MCP servers. Extensions can now inspect or replace MCP tool results before they reach the model.

  8. Anthropic · Claude Code

    Claude Code 2.1.251

    5 new41 fixes25 other

    Added PreModelSwitch and PostModelSwitch hook events (block, confirm, or annotate a model switch); SessionStart resume hooks now receive session staleness and the estimated re-cache cost. Added live streaming of a foreground subagent's tool calls and results to Remote Control clients (background subagents, the default, still show status only).

  9. Anthropic · Claude Code

    Claude Code 2.1.250

    1 other

    Bug fixes and reliability improvements.

  10. Anthropic · Claude Code

    Claude Code 2.1.248

    7 new32 fixes10 other

    Added --restricted (or CLAUDE_CODE_RESTRICTED=1): removes the built-in tools that run commands or code and WebFetch (unless named in --tools), keeps file tools inside the working directory, refuses bypassPermissions, and ignores user, project and local settings files. Added experimental.cacheTtl ("5m" or "1h") to agent frontmatter: a per-agent prompt cache TTL used when no subagent TTL setting is configured.

  11. Anthropic · Claude API

    In Python SDK 1.2.0, TypeScript SDK 0.122.0, Go SDK 1.68.0, Java SDK 2.59.0, Ruby SDK 1.67.0, and C# SDK…

    With this change, client.beta.skills.delete() deletes a Skill together with all of its versions, and the beta Messages type BetaSkill (the container Skill reference) is renamed BetaContainerSkill. Requests that still send the beta headers keep receiving the beta shapes. You can now create personal keys and service account keys in the Claude Console.

  12. OpenAI · Codex app

    Codex app

    Performance improvements and bug fixes. Preserved saved chats and projects when renaming an SSH connection.

  13. OpenAI · Codex CLI

    Codex CLI 0.150.1

    1 fix

    Remote compaction now counts retained images toward its token budget by default, trimming older images as needed.

  14. Google · Antigravity

    Model selection command arguments, reasoning effort autocompletion, transient retry handling, and UI…

    Adds direct /model switching with inline ghost text autocompletion, dynamic /effort hint completion, artifact filesystem event coalescing, and fixes for transient HTTP 502 retries, CPU redraw utilization, and Windows path resolution.

  15. Google · New model

    New model: Gemini Omni Flash

    Gemini Omni Flash generally available (GA): Released gemini-omni-1.1-flash, the GA version of our fast, conversational video generation and editing model. This release includes significant new capabilities: Video extension: Seamlessly extend existing videos by generating continuations at the end of a clip using the extend task or directly with a prompt. Interpolation (first + last frame): Generate a video transitioning between two images using the image_to_video task with up to 2 images. Resolution control: New resolution parameter in video_config supports 360p, 720p (default), 1080p, and 4k outputs. 1080p and 4K outputs are generated using upscaling. The existing gemini-omni-flash-preview endpoint will be deprecated on September 30, 2026. To get started, see the Gemini Omni Flash model page and the omni guide.

  16. Anthropic · Claude Code

    Claude Code 2.1.247

    4 new19 fixes10 other

    Added the SendFeedback tool: when something goes wrong in a session, Claude can draft a feedback report for you to review and send from /feedback (turn off with the feedbackDrafts setting). Added {id, text, cooldownSessions, priority} entries, tipsFile, and label to spinnerTipsOverride, so organizations can rotate their own tips alongside the built-in ones.

  17. Anthropic · Claude API

    The Compliance API session endpoints are out of beta for Cowork and Claude Code sessions

    The Compliance API local session endpoints now also return transcripts of Claude Science sessions (product_surface value claude_science) and Claude for Microsoft 365 sessions in Excel, PowerPoint, Word, and Outlook (product_surface values beginning with office_agents), in beta for Claude Enterprise organizations, with your existing Compliance Access Key and the read:compliance_user_data scope. The Admin API is now available in the ant CLI and the Python, TypeScript, C#, Go, Java, PHP, and Ruby SDKs under client.beta.organization.

  18. OpenAI · Codex app

    ChatGPT for iOS

    Added task search across titles and conversation content on connected hosts. Added a compact composer gauge for viewing and adjusting reasoning effort. Added a full-screen editor for longer prompts. Added configurable Home Screen shortcuts for ChatGPT, Work, and Codex Remote. Added optional comments to selected response annotations. Improved loading for long threads, with older history fetched as needed. Refined task list layout, ordering, and pinning to better match desktop. Inline visualizations now follow iOS appearance and accent settings.

  19. OpenAI · Codex CLI

    Codex CLI 0.150.0

    6 new6 fixes

    Reference other Codex tasks with @ mentions, and ask agents to read, create, or message tasks from the terminal. /copy now offers a picker for full responses, individual code blocks, and blockquotes.

  20. Google · Antigravity

    Voice dictation, embedded ripgrep code search, status line cost metrics, and UTF-8 encoding fixes

    Adds /voice dictation with mic-serve SSH forwarding, embedded ripgrep for consistent agent code search, custom status line cost tracking, granular script runner permission suggestions, and UTF-8 file editing and MCP error fixes.

  21. Google · Antigravity

    Generative UI and UI improvements

    Generative UI renders HTML artifacts inline in chat, with support for KaTeX math and Chart.js and Plotly charts. This release also adds side-by-side terminal splits, the Darcula theme, and @path/to/file references inside AGENTS.md and custom rule files.

  22. Google · New model

    New model: Gemini 3.5 Transcribe

    Gemini 3.5 Transcribe generally available (GA): Released two dedicated speech-to-text models based on Gemini's audio understanding: Gemini 3.5 Transcribe (gemini-3.5-transcribe): High-accuracy, low-latency non-streaming speech-to-text with utterance-based language detection across 85+ languages, speaker diarization, word-level timestamps, and custom vocabulary biasing (up to 1,000 terms). Gemini 3.5 Transcribe Live (gemini-3.5-transcribe-live): Low-latency, bidirectional streaming speech-to-text over WebSockets using the Live API, supporting interim and finalized transcription events, Smart transcription mode, and multiple Voice Activity Detection (VAD) strategies. To get started, see the Audio transcription guide, the Live transcription guide, and the Gemini 3.5 Transcribe model page.

  23. Anthropic · Claude Code

    Claude Code 2.1.246

    3 new48 fixes10 other

    Added a startup warning for Bash allow rules with a wildcard before the subcommand (e.g. Bash(git * main)), since they also match options inserted before the subcommand. Added an Auto mode tab to /permissions for viewing and editing auto mode classifier rules.

  24. Google · Gemini CLI

    Gemini CLI 0.57.0

    1 new7 fixes16 other

    add tool call formatter and integrate failure summaries.

  25. Anthropic · Claude Code

    Claude Code 2.1.245

    1 fix

    Fixed a crash on startup on Linux distributions that ship glibc 2.44 (for example Arch Linux, CachyOS and Fedora Rawhide).

  26. Anthropic · Claude apps

    Memory in Claude Cowork, editable topics, and a sensitive topics setting

    Memory now works across chat and Cowork in the cloud. Everything Claude remembers is listed under Topics in Settings > Memory, where you can edit or delete any item. Topics such as health or beliefs stay out of memory unless you turn on Include sensitive topics in memory. Memory is on by default for Free, Pro, and Max plans and off by default for Team and Enterprise organizations.

  27. OpenAI · Codex app

    Browser extensions, site tools, and cloud sign-in

    More browsers: Use the ChatGPT browser extension in Microsoft Edge, Brave, Opera, and Vivaldi, as well as Chrome. Set up your browser in Settings > Computer Use in the ChatGPT desktop app. All five support tab mentions and browser control; Opera doesn't support side chat. Site tools (WebMCP): In the desktop app's built-in browser, ChatGPT Work and Codex can use tools provided by a website to work with the page. Use GPT-5.6 Sol or GPT-5.6 Terra and update to the latest desktop app. Site tools aren't available with GPT-5.6 Luna or in Enterprise or Edu workspaces. Cloud browser sign-in: On eligible plans, ChatGPT Work on the web, iOS, and Android can ask you to sign in to a supported website through the cloud browser. Enter your details in the sign-in flow, not in the chat. Cloud browser sessions stay separate from your local browser. Website sign-in isn't available for Enterprise or Edu workspaces. Availability depends on rollout and workspace settings. Website-access and action-confirmation requirements still apply. Performance improvements and bug fixes. Fixed browser tab mentions being lost when starting a chat from Home. Improved right-to-left chat titles and Command+Arrow navigation in the composer.

  28. OpenAI · Codex app

    Trigger scheduled tasks from Gmail, Slack, and GitHub events

    ChatGPT scheduled tasks can now run when supported events occur in Gmail, Slack, or GitHub. Filter Gmail messages by sender or subject, watch selected Slack channels, or respond to pull request activity such as reviews, comments, commit updates, and merges. Event-triggered tasks are available in ChatGPT on the web and mobile for eligible plans. Connect the relevant app and approve its requested access before creating a task. The ChatGPT Slack app must be a member of each watched channel, and the connected GitHub app must have access to each watched repository. An event-triggered task can't also use a time-based schedule. When matching events arrive close together, ChatGPT may combine them in one run. Open Scheduled to review pending events or choose Run now. Learn how to trigger scheduled tasks from app events.

  29. Google · Antigravity

    Skill icon branding, empty directory indexing in @-completion, and headless exit code fixes

    Adds skill icon branding with Unicode alignment in catalogs and slash commands, indexes empty directories in @ path completion, auto-approves workspace-scoped read access in review mode, and fixes headless print mode error handling.

  30. Google · Antigravity

    Subagent-scoped custom tools, context compaction lifecycle hooks, and universal JSON schema normalization

    The 0.1.15 release introduces subagent-exclusive tool scoping to reduce context overhead, adds an on_compaction lifecycle hook for observing context checkpointing, expands platform compatibility with Alpine Linux musl wheels, resolves workspace path normalization and custom Vertex endpoint routing, and adds universal JSON Schema normalization for custom Python tools when targeting local OpenAI-compatible LLM endpoints.