Anthropic releases Claude Opus 4.7

#1 📝 Anthropic News

Claude Opus 4.7 - Anthropic launches Claude Opus 4.7, a major upgrade focused on advanced software engineering with improved instruction-following, 3x higher image resolution, and a new xhigh effort level for finer reasoning control. Available at same pricing across Claude products, API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.

#2 📝 OpenAI News

The next evolution of the Agents SDK - Announces the next evolution of the Agents SDK, introducing enhancements to help developers build more capable and reliable autonomous agents and integrations. The post highlights product improvements and new capabilities aimed at accelerating agent development.

#3 𝕏

Sundar Pichai unveiled Gemini on Mac—the first desktop version of the @Geminiapp—built in just days with @Antigravity as a native Swift prototype, with more features on the way.

#4 𝕏

Qwen launched Qwen3.6-Plus and Qwen3.5-Plus in Go with seamless @opencode integration—update now to try out the new models!

#5 𝕏

Google Research unveiled Fabula, an interactive AI writing tool co-designed with 42 expert writers to help authors structure and refine stories. The #CHI2026 demo at 10:30 AM showcases how its convergent iteration approach boosts creativity.

#6 𝕏

Mustafa Suleyman launched MAI Image 2—an efficient new image-generation model—now live on Microsoft Foundry and the MAI Playground.

#7 𝕏

Anthropic co-published a Nature paper on “subliminal learning,” showing that LLMs can covertly inherit and transmit traits—like preferences or misalignment—via hidden signals embedded in training data.

#8 𝕏

Alexandr Wang announces that MSL will publish AI safety and preparedness reports for its frontier models under the new Advanced AI Scaling Framework, starting with the Muse Spark report detailing its readiness measures.

#9 📝 Claude Code Blog

Using Claude Code: session management and 1M context - A how-to for developers using Claude Code that explains session management and working with very large (1M) context windows, with practical guidance and examples to manage state and scale interactions. It covers best practices for keeping sessions efficient and reliable when handling long histories or large inputs.

#10 📝 Simon Willison

Gemini 3.1 Flash TTS - Google released Gemini 3.1 Flash TTS, a text-to-speech model accessible via the Gemini API that outputs audio files and supports director-style prompting. Simon experimented with prompts and built a UI to try it out, showing how voice profile tags and directorial notes can shape the generated speech.

Also covered by: @Simon Willison

#11 𝕏

LlamaIndex 🦙 launched ParseBench, the first document OCR benchmark built for AI agents, and introduced TableRecordMatch (GTRM), a metric that evaluates complex tables as records keyed by column headers.

#12 𝕏

bolt.new shares a video on four GitHub concepts designed to simplify version control and accelerate productivity in your next Bolt.new AI app builder project.

#13 𝕏

Cursor launched interactive canvases for visually representing information. It lets users generate dashboards and custom interfaces that go beyond plain text.

#14 𝕏

Garry Tan shares Part 2 of Resolvers, which acts like an org chart for AI agent sub-tasks. It visually maps and manages multi-step workflows for clearer oversight.

#15 𝕏

Cognition launches Devin in Windsurf—a modular AI agent with dynamic memory, real-time tool chaining, and context-aware planning for faster, more accurate multi-step task execution.

#16 𝕏

Teresa Torres demonstrates how to use git/GitHub, Obsidian Sync, Dropbox, and iCloud to seamlessly sync AI context and skills across devices and teams, tackling absolute-path pitfalls, vault organization by audience, symlink sharing, and each tool’s pros and cons.

#17 𝕏

Tal Raviv likens AI taking over PM tasks to the Mixpanel/Amplitude self-serve analytics boom: while those tools let PMs spin up retention analyses and funnel charts without data-team requests, they also spawned flawed conclusions when events were misinterpreted and expert nua...

#18 𝕏

clem 🤗 is excited to see if autonomous agents can lower the barrier to entry for building open-source AI models and datasets, potentially shifting the balance between closed vs open and off-the-shelf vs customized models.

#19 📝 Eleanor Berger & Isaac Plath

I have given my team access to AI coding tools, but productivity has not improved. Why? - Explores reasons why giving teams access to AI coding tools doesn't automatically raise productivity, focusing on process, expectations, and integration. Offers guidance on adoption practices that actually lead to measurable improvements.

#20 𝕏

Boris Cherny clarifies that the new pricing plans went live six months ago (in November) in response to enterprise customer demand and have been publicly available on their site since then.

#21 𝕏

Philipp Schmid announced the @GeminiApp is now available on Mac, giving macOS users native access to its AI features.

#22 𝕏

Philipp Schmid released Gemini 3.1 Flash TTS today, delivering ultra-fast, high-fidelity text-to-speech generation.

Also covered by: @Simon Willison

#23 𝕏

Mustafa Suleyman introduced MAI-Image-2-Efficient as a production workhorse for high-volume, fast, cost-controlled iterative workflows, and MAI-Image-2 as a precision tool delivering highest-fidelity images with exact details and support for longer, complex text.

#24 𝕏

Cognition launched Windsurf 2.0 with the Devin agent, combining local agents for instant speed boosts and cloud agents that keep working on tasks when you’re away.

#25 𝕏

Garry Tan says OpenClaw now ships with a Docker sandbox, mitmproxy firewall logging, and Clawvisor monitoring, making it fully secure and ready for use. The days of “it’s insecure” for OpenClaw are over.