OpenAI launches GPT-5.6 Luna: health model 25× cheaper

#1 𝕏

OpenAI launched GPT-5.6 Luna, a health-focused upgrade that outperforms GPT-5.5 at its highest reasoning setting while costing 25× less, boosting model quality and making advanced AI more accessible globally.

Also covered by: @There's An AI For That, @AI Explained, @Fireship

#2 𝕏

Sam Altman announces GPT-5.6 as Microsoft 365 Copilot’s new preferred model, powering Word, Excel and Teams with faster, more accurate generative AI.

Also covered by: @There's An AI For That, @AI Explained, @Fireship

#3 𝕏

Google AI launched the Street View grounding feature in Project Genie, letting users generate and explore interactive 360° virtual environments from text prompts or real-world starting points using Google Maps Street View data.

#4 𝕏

OpenAI has relaunched its Bio Bug Bounty as a private, ongoing program with rewards doubled to $50K, inviting AI red teamers and biosecurity experts to hunt for universal jailbreaks against its frontier biology models.

#5 𝕏

Aravind Srinivas launched Computer harness, an agentic orchestration system that natively supports frontier LLMs—Fable, Sol, Opus, Grok, GLM + advisor, Sonnet and GPT-5.5—with subagents across smaller LLMs and multimodal models. Local runtimes are coming soon.

#6 𝕏

Jason Zhou introduced Loopany, a free, open-source loop management space that scaffolds loop contracts, state, logs and programmable triggers. It connects to your team’s local agents to run self-improving cycles from built-in templates.

#7 𝕏

Harrison Chase debuted LangChain this week with NemoClaw DeepAgents, pairing the open-source Deep Agents harness with NVIDIA’s Nemotron 3 Ultra OSS model and the enterprise-ready OpenShell runtime.

#8 𝕏

Sebastian Raschka advises using Luna models with higher-effort settings instead of Sol High or Extra High for agentic coding. He recommends reserving Terra Ultra for peak performance and skipping Sol Ultra’s premium in favor of the Max setup.

#9 𝕏

Alexandr Wang shared Meta AI’s new Coding Agents guide, offering step-by-step instructions, API references, and sample code to help developers build and deploy autonomous coding workflows.

#10 𝕏

Philipp Schmid now lets you choose globally unique ai.studio subdomains for your apps in Google AI Studio (e.g., my-billion-dollar-app.ai.studio). Published apps go live instantly on the web while keeping your code and chat history private.

#11 𝕏

Logan Kilpatrick highlights that you can now deploy apps for free with no billing account required.

#12 𝕏

Cursor launched Side Chats, durable agent threads you can @-mention to ask questions or explore ideas without interrupting your main conversation and seamlessly pull context back into it.

#13 📝 Simon Willison

OpenAI - Quote from OpenAI attempting to clarify ChatGPT Work behavior: web and mobile Work run in the cloud, desktop Work can use local files and remains on the desktop; author notes the clarification was unsuccessful.

#14 𝕏

Rowan Cheung: Meta dropped Muse Spark 1.1, an agentic AI that outperforms OpenAI and Anthropic while massively undercutting their prices—proof that Zuckerberg’s tens-of-billions-dollar compute bet is paying off. Meta’s stock has jumped over 10% since the release.

#15 𝕏

Clem 🤗 highlights HuggingNews, an AI-curated feed by @ivan_bezdomny that cuts through the noise to surface the top AI stories. It’ll soon personalize recommendations via your Hugging Face profile and send you a daily top-10 digest.

#16 𝕏

Google DeepMind digs into neural network interpretability on its latest podcast, where @fryrsquared and @NeelNanda5 unpack chain-of-thought “scratch pads,” mechanistic reverse-engineering techniques, safety auditing methods, and next steps for understanding model reasoning.

#17 𝕏

Santiago warns that using AI as a chatbot delivers just ~0.01% of its potential, while companies like Kulina have used the Viktor agent (integrated in Slack/Teams with access to 3,000+ tools) to scale ad campaigns from 5 to 60 on a $2.

#18 𝕏

Peter Yang suggests treating “Work” and “Codex” as voice-and-tone settings in ChatGPT rather than separate products, and asks if scheduled cloud tasks will be available soon.

#19 𝕏

Garry Tan launched Insforge, a platform where autonomous coding agents bypass clunky APIs and human-focused cloud onboarding to code and deploy on their own.

#20 📝 Claude Code Blog

Working at the frontier: How Cognition trusts Claude Fable 5 to work through the night - A customer story describing how Cognition uses Claude Fable 5 for around-the-clock work, highlighting enterprise AI and coding use cases. The piece illustrates trust and reliability of Claude Fable 5 in production workflows.

#21 ▶️

Grok 4.5 is a bigger deal than Fable 5

Greg Isenberg

#22 𝕏

Cursor launched its Side Chat update with new cloud agent hooks to integrate custom AI workflows, alongside UX and performance enhancements.

#23 𝕏

Peter Yang argues that the ChatGPT app doesn’t need separate Work and Codex brands and should instead use a “Voice: Everyday vs. Technical” distinction within one product.

#24 𝕏

Aravind Srinivas notes that orchestrators drive 60% of a task’s cost and urges using a cost-efficient intelligent model as your orchestrator—even when optimizing for long-horizon agent cost.

#25 𝕏

Harrison Chase notes that Ramp, Stripe, and Coinbase already offer Sierra-like tools. He recommends OpenSWE, a fully open-source, model-agnostic repo with seamless LangSmith integration that they use internally for coding.