OpenAI is having quite a week. GPT-5 turns one tomorrow, and they're celebrating by shipping on two fronts at once.
The pattern is clear: OpenAI is moving from model releases to platform control. Smarter models and the standard everyone must build on. Agent Plugins with Google, Amazon, Microsoft onboard is not a feature, it's a land grab for the agentic layer.
The real question we're watching: who actually owns the ecosystem when the standard is "open"?
|
|
In Partnership with
|
·
|
|
|
|
|
|
Today's Author
Lior Alexander. Founder of AlphaSignal and former ML Engineer. Previously built ML systems at Iguazio, Guesty, Enphase, Mila.
|
|
Top News
|
|
|
|
OpenAI upgrades ChatGPT with GPT-5.6 Sol, cutting factual errors by 68%
|
|
19,428 Likes
|
|
OpenAI just reshuffled what every ChatGPT plan actually runs under the hood. Here is what changed and why it matters. GPT-5.6 Sol now powers all paid chat modes for Plus and Pro users, both quick replies and deep reasoning, as one unified experience. No more switching between modes manually. The accuracy jump is real. 68% fewer factual errors compared to GPT-5.5 Instant, tested across finance, medicine, and law prompts. The model now pushes back when agreeing with you would actually be wrong. Plus and Pro users also get a new reasoning effort slider, so you control how much thinking goes into each response. Fast answer or deep analysis, your call. Free users are not left out. GPT-5.6 Luna becomes the new default with unlimited text chats, plus a Think button that gives Luna extra time on harder questions. - Plus/Pro: updated Sol + reasoning slider, live now
- Free/Go: Luna as default, unlimited text, Think button rolling out
- Codex and Work: unchanged, separate Sol version
|
|
|
|
Presented by Agent Field AI
|
|
Open source hit frontier code review, 10x cheaper
|
|
PR-AF places #2 of 42 on Martian's Code-Review-Bench, ahead of CodeRabbit, Copilot, and Devin, running one open model. 0.706 recall, roughly 3x more valid findings than the commercial tools, at ~10x lower cost per review. It plans a review strategy per PR, runs reviewer agents in parallel, and verifies every finding against your source, so nothing unprovable gets posted. Apache 2.0, self-hosted, runs on any open or closed model. Drops into GitHub Actions.
|
|
|
|
Top News
|
|
|
|
OpenAI launches Agent Plugins open standard to share tools across Codex, Cursor, Copilot
|
|
5,983 Likes
|
|
Until now, building an AI agent plugin meant rebuilding it separately for every tool. Different manifest formats, different folder layouts, different configs. You'd end up maintaining multiple copies of the same thing, watching them slowly drift apart. Agent Plugins 1.0.0 fixes that. It's an open, vendor-neutral standard for packaging your agent skills and MCP servers (think: connectors that let AI agents talk to external tools) into one portable format that works everywhere. At launch, your plugin works across: - ChatGPT, Codex, Cursor, GitHub Copilot, VS Code, and Kiro
- Google is now joining as a core maintainer too
The format is dead simple. A directory, a plugin.json with a name, and your skills folder. That's a valid plugin. Components fail independently, so one broken MCP server won't take down your whole plugin. The steering committee includes Amazon, Cursor, Microsoft, OpenAI, Vercel, and now Google. Build once, ship everywhere.
|
|
|
|
Top Repo
|
|
|
|
Jack Dorsey open-sources a free self-hosted workspace with built-in AI agents
|
|
3,410 Likes
|
|
Jack Dorsey's company Block just open-sourced Buzz, a self-hosted workspace where you and AI agents share the same rooms. Think Slack plus GitHub plus automation, but running on your own server. No subscriptions. No vendor lock-in. 22.8k stars already. Here is what makes it different. Agents are not bots bolted on the side. They join channels, review code, run workflows, and orchestrate other agents, with their own identity and audit trail. Same permissions model as a human teammate. What you can actually do with it: - Host chat, search, Git events, and YAML-based automation in one place, all on your own server
- Add AI agents (Claude, Codex, Goose) to any channel and scope exactly what each one can touch
- Turn a feature branch into a room where patches, CI results, and review all live together
- Query six months of project history in plain English and get answers with receipts
To get started: git clone, run just dev, and your relay is live at localhost:3000. Built in Rust, Apache 2.0 licensed.
|
|
|
At Alpha Signal, our mission is to build a sharp, engaged community focused on AI, machine learning, and cutting-edge language models, helping over 300,000 developers stay informed and ahead. We're passionate about curating the best in AI, from top research and trending technical blogs to expert insights and tailored job opportunities. We keep you connected to the breakthroughs and discussions that matter, so you can stay in the loop without endless searching. We also work closely with partners who value the future of AI, including employers and advertisers who want to reach an audience as passionate about AI as we are.
Our partnerships are based on shared values of ethics, responsibility, and a commitment to building a better world through technology.Privacy is a priority at Alpha Signal. Our Privacy Policy clearly explains how we collect, store, and use your personal and non-personal information. By using our website, you accept these terms, which you can review on our website. This policy applies across all Alpha Signal pages, outlining your rights and how to contact us if you want to adjust the use of your information. We're based in the United States. By using our site, you agree to be governed by U.S. laws.
|
|
|
Looking to promote your company, product, service, or event to 300,000+ AI developers?
|
|
|
|
|