GPT‑6 Cyber announced; Gemini Omni 1.2 Flash API cheaper

#1 📝 OpenAI News

GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents - OpenAI announced GPT‑6 Cyber on November 5, 2026, describing it as a cybersecurity model trained to find and validate exploitable vulnerabilities and as advancing the frontier of aligned cybersecurity agents.

Also covered by: @OpenAI, @Sam Altman

#2 𝕏

Google AI announced Gemini Omni 1.2 Flash, increasing video generation and editing to 15 seconds while improving scene consistency, speech, visual continuity, reference adherence, and cost efficiency. The API price is 30% lower than the previous generation at $0.07 per second for 720p video.

Also covered by: @Google AI, @Logan Kilpatrick, @Demis Hassabis

#3 𝕏

Thinking Machines introduced Inkling 2 Small, which scored 31 on the AA Intelligence Index and is available for fine-tuning and inference on Tinker. A full-weight release is planned soon.

Also covered by: @Thinking Machines, @Mira Murati

#4 𝕏

Personal Finance in Claude is in beta in the US, letting users connect selected accounts and ask money questions in everyday language using their balances and transactions. Access is read-only, so Claude cannot move, manage, or invest users’ money.

Also covered by: @Claude

#5 𝕏

Hugging Face released its open-source ML Intern as an installable Slack bot that can evaluate models, build datasets, and fine-tune models.

#6 𝕏

OpenAI announced plans to expand Daybreak access to organizations of every size, independent researchers, and open-source maintainers. It is also expanding the Daybreak Blue alpha for GPT‑6 Astra, which helps verified defenders validate vulnerabilities and fixes, while Daybreak Red supports deeper exploit development and authorized testing.

#7 𝕏

clem 🤗 reshared @jeffboudier’s post stating that $2m was spent so readers can learn how to train mixture-of-experts models.

#8 🐙 google/skills

google/skills - **google/skills** is Google’s Apache-2.0 repository of reusable Agent Skills for Google products and technologies, covering Google Cloud onboarding, AI/ML, Gemini, Genkit, GKE, infrastructure, and agent management. It has 22,634 GitHub stars and 532 daily trending gains. Worth trying if you need configurable skills for building with Google technologies.

#9 𝕏

clem 🤗 shared @LeRobotHF’s statement that LeRobot v0.7.0 is live, with a video covering what’s new—from terabytes of data to prompting robots.

#10 🐙 The-PR-Agent/pr-agent

The-PR-Agent/pr-agent - **The-PR-Agent/pr-agent** is an MIT-licensed, open-source AI code review agent with GitHub Actions, a local CLI, Docker and self-hosting support, webhooks, docs, and tests. Qodo says it donated the project to the open-source community; the repository has 15,207 stars and gained 1,355 weekly. Worth trying if you need customizable pull-request reviews across multiple Git and AI providers.

#11 🐙 shiyu-coder/Kronos

shiyu-coder/Kronos - shiyu-coder/Kronos is an MIT-licensed family of models for financial K-line forecasting, with model and fine-tuning code, examples, tests, and a web UI. It has 46,999 GitHub stars and 2,979 weekly trending gains. Worth trying if you need customizable OHLCV forecasting across multiple assets.

#12 𝕏

Alexandr Wang said Muse Voice is now available to everyone in the US, enabling phone calls and video chats with Muse. He also said he used Muse for advice about a situation.

#13 𝕏

Ali Ghodsi shared and praised a Databricks Lakeflow Designer demo video showing how the product works, calling it a must-see for Alteryx users.

#14 𝕏

Julien Chaumond retweeted @ThiboNeveu’s statement that 100,000+ 3D objects, including Objaverse assets and CAD parts, were being opened in a unified library for LLM training.

#15 𝕏

Fei-Fei Li shared an invitation to try Atlas in real time, quoting a World Labs post about a Frameboy demo for real-time world generation with the model.

#16 𝕏

Guillermo Rauch shared that Grok and Cursor can now build, deploy, and scale, linking to Vercel’s platforms page while quoting a Cursor AI launch.

Also covered by: @Guillermo Rauch

#17 𝕏

Alexandr Wang announced that Muse launched or became available in Mexico.

#18 𝕏

DeepLearning.AI announced an AI Dev NYC session with Mehdi Jamei and Andi Partovi on designing custom benchmarks through realistic test scenarios, outcome-based measurement, and repeatable evaluations. The event is scheduled for Nov 30–Dec 1, 2026, at Chelsea Piers in New York City.

#19 𝕏

DeepLearning.AI shared enrollment information for its Voice for AI Agents and Applications course with VocalBridge, which explores three ways to add voice to applications and agents. The post features Andrew Ng discussing speed versus reliability and names Ashwyn Sharma in connection with the course.

#20 in

Marc Baselga recommends drafting a release note for every customer-facing roadmap item before committing to build, specifying the customer, benefit, use case, and how to get started. Recapping a Supra Insider discussion with Josh Elman, who spent six years at Apple, Baselga argues that easier AI-assisted building can weaken prioritization and make products harder to use and explain.