News
AI news — page 3
AI news in one feed, newest first: headlines from AI companies in our own words, each linked to its original source, plus selected Codex and Claude posts from Thibault Sottiaux (@thsottiaux), @claudeai and @ClaudeDevs.
Anthropic releases Claude Haiku 5.5
The third Claude 5.5 model is built for high-volume, cost-sensitive work. Anthropic says it costs about 75% less to run than Haiku 4.5, and it halved Sonnet 5.5 cache-read prices on the same day.
Source: Anthropic
Microsoft Execution Containers for AI agents reach general availability
MXC enforces a declared policy of allowed files and network destinations on agents at runtime. Microsoft lists GitHub Copilot and OpenAI Codex among tools that already support it, with Claude Code announced as coming.
Source: Windows Developer Blog
Microsoft pushes local AI on Windows and launches Surface Laptop Ultra
At its Windows event Microsoft described Copilot using local files and local models alongside the cloud, and opened pre-orders for the Nvidia RTX Spark-based Surface Laptop Ultra, from $2,599, available 16 October.
Source: Windows Experience Blog
Google opens its SynthID Detector to everyone
Anyone can now upload an image, video or audio file to check for SynthID watermarks from Google and partner tools. The detector is available globally in English.
Source: Google blog
The community voted for a reset on Day 2, and Tibo says it has been processed.
We shipped four things that were deemed good to great and some math proofs, but the vote is clear and the community demands a reset.…
Excerpt from XDay 2 roundup: free auto-review, simplified builder API, meeting notes and a Decisions API.
Roundup of Day 2/ 2.1/ Approve for me (auto-review) is now included and does not use usage. Can be between 2-10% of plan when used.…
Excerpt from XEvery built a shared company agent on Claude Managed Agents.
The Every team runs as much work as possible through agents. To help the whole team share skills when a new model comes out, they…
Excerpt from XGoogle releases EmbeddingGemma 2 as an open multimodal embedding model
The roughly 740M-parameter model maps text, code, images, video and audio into one embedding space and ships under Apache 2.0 on Hugging Face and Kaggle.
Source: Google blog
Artificial Analysis: Mistral Large 4 is the top model from outside the US and China
The preview scored 38 on the Artificial Analysis Intelligence Index, close to GPT-6 Luna and DeepSeek V4.1 Flash at max effort, and 50 on its Cyber Index.
Source: Artificial Analysis
Claude now works inside Google Docs, Sheets and Slides.
Claude now works inside Google Docs, Sheets, and Slides, and those files also open inside Claude. In Google Workspace, Claude sits in a sidebar next…
Excerpt from XThe Claude Startups program expands to more founders.
The Claude Startups program is expanding to more founders. Members can get a year of Claude Team, API credits, special offers on the tools a…
Excerpt from XClaude comes to Google Docs, Sheets and Slides
A Claude for Google Workspace add-on is in public beta on paid Claude plans, alongside new beta connectors that let Claude create and edit Google files directly.
Source: Claude (Anthropic)
Mistral previews Mistral Large 4, with open weights due this month
Mistral's largest model so far is a natively multimodal mixture-of-experts model with 1 trillion total parameters. A preview API is live in Mistral Studio, and Mistral says the weights will be released by the end of October.
Source: Mistral AI
OpenAI publishes math results from an unreleased internal model
OpenAI released a set of mathematical results in a GitHub repository, with Lean formalizations for many proofs, and says it is working to release the model that produced them responsibly.
Source: OpenAI
Day 2.1: auto-review becomes free for users signed in with a ChatGPT account.
Day 2.1/ We have made Auto-review free for all users signed in through a ChatGPT account. You can enable it in settings > permissions >…
Excerpt from XAlso tracked: Usage reset tracker · All releases · Blog