New AI models

Every day we read hundreds of stories. Only the most important ones are here — short, in our own words, with a link to the original source.

1. New AI models

   Anthropic releases Fable 5.2 and Mythos 5.2

   Anthropic has launched Fable 5.2, offering faster output and higher benchmark scores, alongside Mythos 5.2. Mythos 5.2 remains limited to verified groups.

   Full story at TestingCatalog AI News ↗
2. New AI models · Artificial intelligence

   Google launches Gemini 4 Carbon

   Google is making Gemini 4 Carbon generally available after a two-week test, topping benchmarks ahead of Anthropic's Fable 5.2 at lower cost. It also previews a '/swarm' feature for coordinated AI agent teams.

   Full story at The Decoder ↗
3. Artificial intelligence · New AI models

   Fable 5.2 completes 41 percent of real freelance jobs

   Anthropic's Fable 5.2 completes 41.3 percent of real freelance jobs at professional quality on the Remote Labor Index. A year ago the top score was 2.5 percent, and product design and animation remain weak spots.

   Full story at The Decoder ↗
4. New AI models · Artificial intelligence · Cybersecurity

   Anthropic releases Claude Fable 5.2 and Mythos 5.2

   Anthropic's most powerful models come in a free version and one for verified organizations. The company says they solve open physics problems and write over 90 percent of merged core code changes, with self-accelerating capabilities deliberately curbed.

   Full story at The Decoder ↗
5. New AI models · Artificial intelligence

   Gemini 4 Carbon arrives in Google Search AI Mode

   Google released Gemini 4 Carbon, its strongest model with frontier performance in cybersecurity and long-horizon agentic workflows. It is available to subscribers in Google Search via AI Mode.

   Full story at seroundtable.com ↗
6. New AI models · Artificial intelligence · Chips & hardware

   Nvidia's GR00T 2 teaches robots from human videos

   Nvidia's GR00T 2 transfers human hand movements from videos to robotic arms, trained on about 400,000 hours of footage. Success rates range from 10 to 80 percent depending on the task.

   Full story at The Decoder ↗
7. Artificial intelligence · New AI models · IT industry & companies

   Gemini Robotics 2.5 sorts parts in a Mercedes pilot

   Google Deepmind introduced Gemini Robotics 2.5, a new control model for robots. Tested with Apptronik on Mercedes-Benz lines in Hungary and Germany, the Apollo 2 robot sorted metal parts with a reported 96 percent success rate.

   Full story at The Decoder ↗
8. Artificial intelligence · New AI models · Cybersecurity · Programming & dev tools

   DeepMind unveils AlphaProof Code for verification

   Google DeepMind introduced AlphaProof Code. Its agents wrote three million lines of proofs in Lean to verify Signal's core encryption.

   Full story at Google DeepMind ↗
9. Cloud & data centers · Artificial intelligence · New AI models · Enterprise software

   Google Cloud announces new AI offerings

   Google Cloud made Gemini 4 Argon available for complex enterprise work. It also announced general availability of an Anthropic-managed deployment option for Claude models.

   Full story at Analytics Insight ↗
10. Artificial intelligence · New AI models · Programming & dev tools

    Meta introduces Muse Ember model

    Meta launched Muse Ember for low-cost agent and coding tasks. It offers multimodal reasoning, a 1M-token context window and low API pricing.

    Full story at TestingCatalog AI News ↗

An AI writes the short texts from the source's stories.