U.S. chatbot use passes 50 percent: AA-Briefcase benchmark measures knowledge work
ARD, an open spec for discovery. North Mini Code gains traction. Security experts criticize U.S. government. Apple Intelligence, beyond Siri.
Latest
Reinforcement Learning With Hints: Privileged On-Policy Exploration (POPE) trains models to expand on partial solutions
Reinforcement learning can’t train a model to solve a difficult problem if the model doesn’t discover all the right steps.
Nvidia’s Nemotron Goes Big: Nvidia Nemotron 3 Ultra bets on speed and openness to win customers
Nvidia’s largest-yet model is among the best-performing from a developer based in the U.S. and among the most open developed by anyone.
Agentic Tests Beyond the Bug Hunt: DeepSWE, ProgramBench, and ITBench-AA push agents harder than SWE-bench
SWE-bench, a family of benchmarks that focuses on an LLM’s ability to fix software bugs, is giving way to new tests that evaluate agent software-engineering performance in more challenging ways.
Claude Fable 5’s Benchmark Problems: Independent tests of Claude Fable 5 run into Anthropic's protective policies
Before Anthropic pulled its latest Claude models from circulation, even professional testers couldn’t readily tell whether they were getting a Mythos-class model or a lesser version under the same name.
Open Platforms Beat Power Plays
Over the last two weeks, both the U.S. Government and Anthropic took significant actions that demonstrated their power to control access to AI by restricting what others can do with frontier models.
Testing Mythos and Fable, Moving Beyond SWE-bench, Nvidia's Open Contender
The Batch AI News and Insights: Over the last two weeks, both the U.S. Government and Anthropic took significant actions that demonstrated their power to control access to AI by restricting what others can do with frontier models.
Zhipu’s GLM-5.2 is the new top open model: SpaceX buys coding favorite Cursor
OpenRouter’s model mix-and-match. Subject expertise trumps software skills. OpenAI loses share to Google, Anthropic. Google ruled liable for AI mistakes.
DiffusionGemma breaks 1,000 tokens/second: Anthropic closes Fable and Mythos after US intervention
Claude Fable 5 no longer silently degrades. Hermes Agent maker streamlines setup. Agents’ Last Exam pushes top models. Gemini-SQL2 translates database queries.
State Media Influences LLM Responses: Significant portions of AI training material reflect national propaganda
Popular large language models have adopted the biases of governments that control the free flow of information, particularly when those models generate output in the languages of countries where such governments are in power, researchers found.
RSI Is the New AGI: What Is recursive self-improvement, and why Is everybody talking about it?
The phrase recursive self-improvement erupted on social media following an Anthropic report that tracked AI-driven gains in the company’s internal software-engineering productivity.
Cursor Fits Its Model to Its Agent: Composer 2.5 for Cursor rivals GPT-5.5's coding abilities at lower price
Cursor’s latest software engineering model rivals the performance of leading competitors like Claude Opus 4.7 and GPT 5.5 for a fraction of the price.
Machine Learning Research
Behold Mythos!: Anthropic released Claude Mythos 5 and Claude Fable 5, a public version with safeguards
After months of headlines that teased a large language model with extraordinary capabilities, Anthropic launched Claude Mythos 5, which can crack software previously believed to be secure, and Claude Fable 5, a version for general use that limits what users can do in an unprecedented way.
Letters
Agents on the Desktop
If you haven’t already, I encourage you to experiment with using AI agents not just to chat but to actually do work for you on your desktop.
The Batch Newsletter
Mythos Begets Fable, Cursor's Composer 2.5, Agents Building Agents
The Batch AI News and Insights: If you haven’t already, I encourage you to experiment with using AI agents not just to chat but to actually do work for you on your desktop.
Data Points
Claude Fable 5, or Mythos for the masses: Apple rethinks mixture-of-experts to save local memory
Google’s voice translation model covers 70+ languages. OpenAI’s preliminary public-offering paperwork. NotebookLM, now powered by Gemini 3.5 agents. FrontierCode, a new code-quality benchmark from Cognition.
Data Points
Apple builds local/cloud models with Google: Gemma 4 12B, a laptop-sized model with multimodal power
The first working vaccine built by AI. Kimi CLI, Moonshot’s software engineering agent. The White House’s plans for an OpenAI stake. OpenJarvis, an open-source agent that learns on-device.
Machine Learning Research
Fine-Tuning LLMs to Expand on Summaries Unearths Pretraining Texts: Fine-Tuning can strip models of copyright alignment guidelines
Fine-tuning large language models on a seemingly benign task that would be useful to writers — expanding plot summaries into paragraphs of polished fiction — causes them to regurgitate substantial portions of books on which they were pretrained.
Tech & Society
Inside the Gray Market for LLM Access: Middlemen package extra tokens, hijack IDs to resell, distill models
An ecosystem of API proxy servers enables AI developers in China to access top U.S. models at deeply discounted prices.
Tech & Society
How AI is Saving Whales: WhaleSpotter pairs sensors with AI algorithms to detect marine mammals
An AI-powered network of thermal sensors is helping ships avoid collisions with whales.
Machine Learning Research
Qwen3.7-Max Adds Speed and Power: Alibaba's latest proprietary model challenges U.S. rivals
Alibaba updated its flagship large language model for long-running agentic work, pushing it into the top rank among LLMs built in China.
Letters
AI Regulations Must Balance Innovation and Risk
There have been intense efforts over the past few years to lobby governments to pass AI laws for regulatory capture or to suppress open source.
The Batch Newsletter
Qwen3.7-Max Challenges Google for Third Place, AI Saves Whales, Fine-Tuning Breaks Copyright Alignment
The Batch AI News and Insights: There have been intense efforts over the past few years to lobby governments to pass AI laws for regulatory capture or to suppress open source.
Data Points
Microsoft fully trains its own models: Copilot app brings agent management to desktop
How agents think about search. Hermes now a multi-platform desktop app. Qwen3.7-Plus, Alibaba’s midsized cloud model. OpenAI’s latest plugins for Codex.