Latest updates

Released September 23, 2026

Key changes

• Rollouts and Security Review
• Rollouts
• Security Review
• Get started

Full changelog →

Released October 1, 2026

Key changes

• Added Claude Mods: plugins may now modify deeper behavior
• Added You should know, a built-in mod where a side agent watches your back and flags things you or Claude might miss. Turn it on with /plugin enable cc-plugin-you-should-know@builtin (for first-party sessions with…
• Added an n:<text> filter to the agents view that matches session names and tasks; a filter now shows matches in collapsed sections and Enter opens the first match
• Added prompttext to the OpenTelemetry userprompt event, a copy of prompt for backends that nest dotted keys; drop or mask it wherever you drop or mask prompt (anthropics/claude-code#70763)
• Added URL prompts from MCP servers on the 2025-11-25 protocol, for example to sign in. If a server no longer connects after this update, add "bareElicitationCapability": true to its MCP config entry
• [VSCode] Added "Run in background" to a running command or sub-agent, to move it to the background and keep working

Recent releases

• v2.1.286 · September 30, 2026

  Added a count such as "2 of 5" to the permission prompt when several permission requests stack up

  Added mouse support for the "N more" rows of lists in fullscreen mode: click one to jump to that end of the list, with hover and pressed states

  [VSCode] Added bookmarks: save Claude's responses and keep them in view in a Bookmarks side panel

• v2.1.285 · September 29, 2026

  Added CLAUDECODEDISABLEWEBFETCH environment variable to turn off the WebFetch tool

  Added claude --desktop to open the Claude desktop app on the current directory, or on a session with --continue / --resume <id>

  Added claude plugin configure <plugin> to show a plugin's options and which are unset, or save new values read from stdin with --values-stdin

• v2.1.284 · September 28, 2026

  Added Claude Sonnet 5.5 (claude-sonnet-5-5), now the default Sonnet model on the Anthropic API — 1M context, $2/$10 per Mtok with $0.20/Mtok cache reads

  Added a "Yes, but ask again next time" answer to auto mode's prompt before a read outside the working directories, so you can allow that one read and still be asked about later ones

  Added dollar amounts to the Claude apps gateway spend limit in /usage and the status line (for example "$271.40 / $500.00 spent this month") when the gateway runs this version or later; the status line's…

• v2.1.283 · September 25, 2026

  Added x-claude-code-prompt-id to the gateway hint headers so LLM gateways can group the requests that serve one user prompt; opt in with CLAUDECODEGATEWAYHINTHEADERS=1

  Added availableModelsMatch managed setting: with "exact", an availableModels entry allows only the model version it names, so new releases stay blocked until listed

  Added deniedModels managed setting to block specific models, even when availableModels allows them

Full changelog →

Released October 8, 2026

Key changes

• Voxtral Mini 4B Realtime Arabic is a streaming speech-to-text model for Arabic dialects and Modern Standard Arabic. It is designed to handle challenging real-world audio and code-switching, where speakers alternate between languages within the same conversation.

Recent releases

• Mistral Large 4 1T A52B · September 30, 2026

  Mistral’s largest model so far: a natively multimodal Mixture-of-Experts model with about 1T total and 52B active parameters, and a 1M-token context window.

• Shieldstral 1.0 3B · July 16, 2026

  Shieldstral is a compact 3B-parameter, policy-adaptive multimodal safety classifier. Instead of predicting a fixed set of moderation categories, Shieldstral evaluates content against a safety policy expressed in natural language and returns a single continuous safety score. This makes it a flexible drop-in guardrail…

• Leanstral 1.5 119B A6B · July 1, 2026

  Leanstral 1.5 is an open-source code agent model designed for Lean 4, a proof assistant capable of expressing complex mathematical objects such as perfectoid spaces and software specifications like properties of Rust fragments.

• Mistral Medium 3.5 128B · March 31, 2026

  Mistral Medium 3.5 is our first flagship merged model. It is a dense 128B model with a 256k context window, handling instruction-following, reasoning, and coding in a single set of weights. Mistral Medium 3.5 replaces its predecessor Mistral Medium 3.1 and Magistral in Le Chat. It also replaces Devstral 2 in our…

All releases on Hugging Face →

User guide and local install →

Released June 13, 2026

Key changes

• Open-weight, natively multimodal model (text and images) with a 1-million-token context window. Moonshot calls it its most capable model to date and “the world’s first open 3T-class model”.
• Mixture-of-experts: 2.8T total parameters, 104B active, 16 of 896 experts selected per token. Built on Kimi Delta Attention and Attention Residuals, which Moonshot says scale about 2.5× more efficiently than Kimi K2.
• Trained with MXFP4 weights and MXFP8 activations from the SFT stage on, so the released weights are already quantized.
• API id kimi-k3 on platform.kimi.ai, with OpenAI- and Anthropic-compatible endpoints. Self-hosting is supported on vLLM, SGLang and TokenSpeed.
• Thinking is always on: reasoning_effort takes low, high or max (default max). In multi-turn and tool-use chats you must send back the full assistant message, including reasoning_content and tool_calls.

Recent releases

• Kimi K2.7 Code · June 11, 2026

  Kimi K2.7 Code is a coding-focused agentic model built upon Kimi K2.6. With substantial improvements on real-world long-horizon coding tasks, it strengthens end-to-end task completion across complex software engineering workflows while improving token efficiency, reducing thinking-token usage by approximately 30%…

• Kimi K2.6 · April 14, 2026

  Kimi K2.6 is an open-source, native multimodal agentic model that advances practical capabilities in long-horizon coding, coding-driven design, proactive autonomous execution, and swarm-based task orchestration.

• Kimi K2.5 · January 1, 2026

  Kimi K2.5 is an open-source, native multimodal agentic model built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base. It seamlessly integrates vision and language understanding with advanced agentic capabilities, instant and thinking modes, as well as…

• Kimi K2 Thinking · November 4, 2025

  Kimi K2 Thinking is the latest, most capable version of open-source thinking model. Starting with Kimi K2, we built it as a thinking agent that reasons step-by-step while dynamically invoking tools. It sets a new state-of-the-art on Humanity's Last Exam (HLE), BrowseComp, and other benchmarks by dramatically scaling…

All releases on Hugging Face →

Released October 17, 2026

Key changes

• A minor update to DeepSeek-V4.1-Flash. DeepSeek reports better reasoning, coding, agentic and visual scores, with shorter output and less task execution time.
• DeepSeek’s own results against V4.1 Flash: Terminal-Bench 4.0 42.3 (from 31.2), DeepSWE v1.1 77.0 (from 74.2), ProgramBench 29.1 (from 20.3).
• Keeps the 1M-token context window, and DeepSeek recommends max_tokens of at least 256K. Weights are under the MIT License.

Recent releases

• DeepSeek V4.1 Flash · September 10, 2026

  Multimodal Mixture-of-Experts model with 552B backbone parameters and contexts of up to 1M tokens.

• DeepSeek V4 Flash Vision Exp · August 31, 2026

  We are excited to introduce DeepSeek-V4-Flash-Vision-Exp, our first experimental multimodal model in the DeepSeek-V4 family. It builds on the DeepSeek-V4-Flash architecture by incorporating visual modules and undergoing continued training to unlock visual understanding capabilities.

• DeepSeek V4 Pro 0813 · August 13, 2026

  DeepSeek-V4-Pro-0813 is the official release of DeepSeek-V4-Pro, superseding the preview version, with greatly enhanced agentic capabilities and performance improvements that are especially pronounced in production environments. It is built on the DeepSeek-V4-Pro (Preview) model structure, with a DSpark speculative…

• DeepSeek V4 Flash 0731 · July 31, 2026

  DeepSeek-V4-Flash-0731 is the official release of DeepSeek-V4-Flash, superseding the preview version, with substantially enhanced agentic capabilities. It has the same model structure as DeepSeek-V4-Flash-DSpark, i.e. it comes with a speculative decoding module attached.

All releases on Hugging Face →

Featured and recent listings from the official catalog

• dr eggbot — Lauren Tan. Designs high-quality Grok Bots. Asks a few preference questions, then creates them with CreateAgent. Coding bots get the poteto-mode bar (one job, unslopped, verified). Non-coding bots get the same tightness: one job,…
• Overheard — Lenny Rachitsky. Watches Reddit, Hacker News, news sites, and X for third-party mentions of your name, brand, and URLs, then sends a short weekday digest when something clears the bar. Stays quiet on dead days and never posts on your…
• Tradbot — Claire Vo. Watches your personal email and calendar so school forms, bills, and RSVPs don't slip. Drafts the reply, catches the pickup clash, and never sends without you.
• Projects Manager — Eric Zakariasson. Runs your team's projects from Notion: one row per project, a channel per project, and tasks your specialist bots claim. You decide, agents execute, and it never does the specialist work itself.

Grok Bot guide →

Anthropic · OpenAI · xAI · Microsoft · GitHub Copilot · Mistral · Moonshot · DeepSeek

• MAI Image 2.7 debuts above GPT Image 2.5 Flare on Artificial Analysis and #3 on Arena — Microsoft, October 27, 2026
• Introducing GPT-6.1 Terra — OpenAI, October 26, 2026
• GPT-6.1 Terra in GitHub Copilot — GitHub Copilot, October 26, 2026
• DeepSeek V4.1 Flash 1017 — DeepSeek, October 17, 2026
• Grok Voice 3.0 is available — xAI, October 14, 2026

What’s new in each model →