Anthropic Launches Claude Fable 5.2 and Mythos 5.2
Anthropic’s new Claude models generate tokens up to 30% faster and cost less than their predecessors. Fable reaches paid plans, while Mythos stays restricted.
Comprehensive coverage of academic breakthroughs, patents, labs and experimental models pushing the frontiers of technology. Discover how cutting-edge research turns into practical innovation across industries and influences the next generation of products. This category captures the pipeline from idea to market.
Anthropic’s new Claude models generate tokens up to 30% faster and cost less than their predecessors. Fable reaches paid plans, while Mythos stays restricted.
ActivTrak found workday spans of 8.4–9.1 hours for AI users versus 7.6 for non-users. The gap does not prove AI caused longer working hours.
OpenAI’s Greg Brockman expects major AI health gains within 12–18 months and says death could one day become a choice, without setting a timeline.
Anthropic says future Claude models could sharply accelerate AI research, leaving less time to evaluate new capabilities and put safeguards in place.
MIT researchers used GPT-6 Astra to edit Qwen’s weights without retraining, lifting MATH-500 from 7.9% to 48.1% by reducing reasoning loops.
Claude Fable 5.1 leads HalachaBench’s Jewish law test at 79.7/100. The AI-graded results are provisional and await rabbinic validation.
SuperMatics is using AI to search for superconductors that can become commercial wire, with AI power infrastructure and fusion magnets among its targets.
Nine of ten models raised wallet limits in at least one trial; five paid an impersonator when tasked only with summarizing email.
Anthropic details Claude-assisted cancer research and its own rare-disease programs, expanding its biotech role alongside pharmaceutical partnerships.
Google’s Genie 4 adds sound, dialogue and live prompt controls to AI-generated worlds, with a robot simulation demo and wider access still to come.
Russian text used about 42% more tokens in one test, while a wider study found a 17x gap between languages. Results depend on the tokenizer.
OpenAI and Anthropic claim AI-generated proofs refute the Exponential Time Hypothesis. Lean checks support the claims, while expert scrutiny remains essential.