News
Models & Labs
New models, releases and benchmarks from OpenAI, Anthropic, Google, Meta, Mistral and the open-weight world.
631 stories · Every headline links to the original publisher. HapnGo curates and groups; we do not copy articles.
Paris-based Mistral releases "Le Chonk," a strong coding model that still trails China overall
Mistral has released the weights for Large 4, aka "Le Chonk," a trillion-parameter model trained entirely in Europe. Mistral says…
Mistral releases weights for its trillion-parameter AI model, targeting cybersecurity
Mistral says its new model can afford organizations a kind of control over its cyber defenses that closed models can't.
Mistral Says Its AI Model ‘Le Chonk’ Has Leapfrogged China in the Open-Weight Race
The French AI lab claims its latest release matches or outperforms even the best models from DeepSeek and Z.ai in coding and…
Anthropic is asking Wall Street to believe in AGI
Today on Decoder, I’m joined by Joey Brookhart, an analyst at the chip and AI research firm SemiAnalysis, to help me dig into the…
Chinese spies targeted Nvidia CEO Jensen Huang, according to The New York Times
AI summary US intelligence believes Alice Huang (Huang Le) of Singapore data center firm Megaspeed, who is close to Nvidia chief Jensen Huang, feeds information to Beijing, per The New York Times. Trump was told roughly a year ago but took no action, and the Nvidia CEO was never warned. Washington also suspects Megaspeed helps China obtain Nvidia chips; a Malaysian affiliate bought at least $4.8 billion in AI hardware from June 2024 to February 2026. Nvidia says Megaspeed gets no sensitive information, Megaspeed has opened an internal probe, and China's embassy rejected the claims.
Anthropic’s business plan: Build God or die trying
The Anthropic S-1 seems to have provided the financial justification for the company to spend as much as it possibly can on…
OpenAI's Astra becomes the main target for Chinese AI labs after Anthropic tightens its defenses
AI summary DeepSeek and Moonshot AI reportedly relied heavily on outputs from OpenAI's GPT-6 Astra to train recent models, including V4.1-Flash 1017 and Kimi K3.1 Pro, according to people familiar with the matter cited by The Information. Such distillation requires coaxing models into exposing hidden reasoning, which both OpenAI and Anthropic prohibit. Anthropic's Claude was previously the favored target, but the company says new safeguards reduce extractable data by 90 to 99 percent. OpenAI, which flagged a coordinated campaign tied to Moonshot-linked people in late September, says attackers have adapted after each new defense.
Google AI Mode Tests Blending All Products With Sponsored Products
AI summary Google is running an experiment in AI Mode that places paid Shopping Ads next to unpaid visual product results inside carousels. A spokesperson confirmed the test, saying such ads show up for commercially relevant searches and that every paid listing carries a "Sponsored" label. Sachin Patel first noticed the format and shared screenshots on SEW. Google began blending ads with organic results in October 2023, a practice it calls dynamic ad placement.
How we detect frontier model misalignment
OpenAI uses layered monitoring to detect frontier model misalignment and unauthorized agent behavior so responders can intervene…
Has anyone gotten BiRefNet to work on mobile NPUs and GPUs? especially Qualcomm [D]
I've been trying to get BiRefNet to work on Qualcomm's mobile GPUs and NPUs. But i guess it's deformable convolutions are pretty…
New AA intelligence index v5 with “Private tests” conveniently upranks GPT, Google’s models, while open weights chinese models downrank
The new Artificial Analysis intelligence index v5, is being criticised, because its introduction of "trust me bro" private tests…
How Ropes & Gray is developing AI for transactional diligence
Ropes & Gray is developing AI for transactional diligence with OpenAI. Lawyers review grouped documents, summaries, and reports…
Polylane for ChatGPT
Ask ChatGPT what broke in production Discussion | Link
OpenAI reportedly plans to buy Google's AI chips (TPUs) — rivals becoming each other's suppliers
Reuters reported on Tuesday (Oct 28), citing The Information and a person familiar with the matter, that OpenAI plans to buy…
Floot in ChatGPT
AI summary Floot, an app builder already used over MCP with Claude and ChatGPT, is now available as a native app in the ChatGPT sidebar. It shows live previews in the chat, reuses context from the conversation, and can publish apps to a web link, Google Play and the App Store, with database, logins and hosting included. Users build on their existing ChatGPT plan, Floot does not charge for AI tokens, and there are 100 free build actions per day. It is installed from ChatGPT's app directory and invoked by typing @Floot.
Anthropic just launched tools to show companies what their AI spending is buying
The frontier lab is lifting the curtain on how users spend money on its models ahead of a planned IPO.
Google Deepmind releases Gemini 4 Carbon just two weeks after Argon with gains across the board
Just two weeks after Gemini 4 Argon, Google DeepMind is releasing Gemini 4 Carbon, which tops the Artificial Analysis…
Free, daily
The HapnGo Briefing, in your inbox
One email a day with the AI stories that matter, why they matter, and one tool worth trying. No spam, one-click unsubscribe.