Anthropic built internal Scion after monorepo merges passed 30,000 a day
Anthropic migrated its monorepo from git to internal Scion after monorepo merges jumped from about 2,500 a day to more than 30,000.
News
OpenClaw 2026.10.3 adds Muse Wonder and a MiniMax preview
OpenClaw 2026.10.3 lets Meta users pick Muse Wonder and adds MiniMax M3.1 Flash Preview, plus update and recovery fixes.
Gemini 4 Argon is live on the paid API and for Ultra in Antigravity
Gemini 4 Argon is open on the paid Gemini API and for Ultra users in Antigravity, at an introductory API price.
Claude Code 2.1.299 compacts idle terminal sessions before the prompt cache expires
Claude Code 2.1.299 compacts large idle terminal chats before the prompt cache expires and can pass @ mentions as paths.
Meta ships Muse Wonder and keeps Muse Wonder Full inside Muse Code
Meta releases Muse Wonder and Muse Wonder Full for agent tasks. Wonder is on the Model API and in Muse Code; Full stays with Muse Code subscribers.
The X Subscription puts Grok, Cursor, and X Premium on one bill
SpaceXSI puts Grok, Cursor, and X Premium on one optional plan, The X Subscription, from $8 to $200 a month with a shared weekly usage pool.
OpenCode 2.0 puts every client on one local service
OpenCode 2.0 runs terminal, desktop, and web against one local service, rebuilds the client API, and adds Code Mode plus an embeddable SDK.
Cloudflare's API MCP server now serves Cloudflare skills
Cloudflare's API MCP server now serves its skills to clients that support the Skills over MCP extension, via skills/list and skill:// URLs.
A Business Insider workplace test hit ChatGPT Dot sharing and login walls
Business Insider’s October 10 account shows a workplace Dot blocked by Enterprise sharing, then useful after he got approval to connect a newsletter tool.
A ChatGPT dot can now edit Work automations and pick up Codex threads
OpenAI’s October 9, 2026 note lets a ChatGPT dot start Codex threads and edit existing Work automations.
Anthropic says Claude took unintended actions on real sites in evaluations and internal use
Anthropic's Oct. 9 report details four kinds of unintended Claude actions on real sites and a live-internet cutoff for internal evaluations.
Nace open-sources Drex 1.5, a model that scores choices
Nace open-sourced Drex 1.5, a small model that scores supplied choices instead of writing text, with local runtimes.
Google Cloud’s Gemini agent is one workplace agent, still in private preview
Google Cloud’s Gemini agent plans workplace tasks from one prompt, in private preview, and can route jobs across Gemini and Claude.
GPT-6.1 Sol Ultrafast is live on Amazon Bedrock
AWS opened Ultrafast mode for GPT-6.1 Sol on Amazon Bedrock, a six-times-priced tier for latency-sensitive agent calls.
Sierra publishes the Poppy draft and 35 more design partners
Sierra published the Poppy draft of Personal Agent Protocol and named 35 more design partners, including OpenAI, Visa, and Cloudflare.
Claude Code lets in every Pro and Max user already on the Projects waitlist
ClaudeDevs says it let in every Pro and Max user already on the Claude Code Projects waitlist. The beta is still rolling out.
Asana's browser-agent tests got 76x cheaper on GPT-6.1 Sol
Asana and OpenAI say a cache-preserving browser-agent workflow on GPT-6.1 Sol was about 76x cheaper and 5x faster in tests.
OpenAI says Sophos Daybreak agents cut handled-case response to 89 seconds
OpenAI says Sophos Daybreak agents cut handled-case response from about 38 minutes to 89 seconds, with analysts still in the loop.
Claude Code 2.1.296 lets subagents compact earlier and pins workflow models
Claude Code 2.1.296 adds a per-subagent auto-compact window, a shared model for workflow agents, and a managed code policy in Desktop.
Claude Startups pauses the free year of Claude Team and $1,000 API credit offers
Anthropic paused the Claude Startups free year of Claude Team and the $1,000 API credit offers after a surge of applications. All applications are being re-reviewed; an already-claimed offer does not change, and some application statuses may change.
Claude Docs, Slides, and Design leave beta as Dashboards and Motion stay limited
Claude Docs, Slides, and Design are generally available on every plan, while live dashboards and Motion animations stay in beta.
Anthropic opens OSS Scanner, an opt-in service that sends unreviewed model bug reports
On Oct. 8, 2026, Anthropic launched OSS Scanner, a free opt-in service that sends eligible open-source projects fully model-generated vulnerability reports, including reproducers and, when available, candidate patches, without human review.
Nous raises $90 million and takes Hermes to businesses
Nous Research raised $90 million and is launching Hermes for Businesses. TechCrunch reported a $1.5 billion Series B valuation.
Claude Code 2.1.295 fails closed when a hook cannot run
Claude Code 2.1.295 blocks an action when a command or HTTP hook fails to start, and lets terminals show working, waiting, or done.
Sonnet 5.5 cache reads drop to $0.10 per million
Sonnet 5.5 cache reads now cost $0.10 per million tokens on the Claude Platform API, with Claude Code usage limits unchanged.