Claude Fable 5.2 and Claude Mythos 5.2 are upgraded versions of the same model, built for demanding technical work. The announcement highlights advances in coding, reasoning, vision, science, engineering, and robotics, with examples from prerelease testing including a Python-to-Rust service migration, three working electronic devices designed from scratch, progress on black-hole physics problems, and a formalization of Fermat’s Last Theorem. In one experiment, Fable coordinated more than 1,000 subagents over eight days to build an operating system that boots on a laptop. The models produce answers up to 30% faster than version 5.1 and cost $8 per million input tokens and $40 per million output tokens; an optional mode boosts generation speed up to 2.5 times at double the price. For routine coding and knowledge tasks, Opus 5.5 remains the recommended starting point.
Benchmark results generally place Fable 5.2 ahead of Opus 5.5, with especially large gains in robotics, agentic coding, science research, and computer use. It scores 49.1% on FrontierCode at its highest effort, versus 45.5% for Opus, and often matches Opus’s best results at substantially lower cost. Mythos 5.2 leads the listed cybersecurity and protein-design tests, though access to it is restricted to qualifying research organizations. Fable is available broadly with safeguards against cyber and biological misuse. Prerelease safety evaluations found alignment performance comparable to or better than recent models, while safeguards reportedly block fewer legitimate requests than before; additional protections target attempts to distill the model. Fable can also advise other models or coordinate smaller models on complex tasks.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Researchers tested 16 LLMs on all 55 TIS-100 puzzles, giving each one attempt and time to refine its code. The models found 133 new performance tradeoffs across 18 puzzles and set three outright cycle records, though human players improved on most solutions within days. (tis-100.wincon.dev)
Australia is considering a new criminal offence for AI companies that fail to take reasonable steps to prevent harm caused by their models. The government is also weighing an AI safety watchdog with power to impose significant fines, following a hack of a Medicare statistics portal involving OpenAI. (abc.net.au)
AI is increasingly becoming a gateway through which people find information. Audiences still resist having AI produce their news. (niemanlab.org)
Nvidia holds a central position in the AI industry, making disruption to the company potentially catastrophic for the wider ecosystem. The race for advanced AI is also intensifying economic and geopolitical conflict, while concentrated corporate power and infrastructure face growing political resistance. (france24.com)
LLM inference can require reading about 26 GB of model weights to generate each token from a 13-billion-parameter model. The explanation shows how model size drives memory bandwidth demands during text generation. (youtube.com)
Deletion Tuning matches full fine-tuning across three LLM post-training methods by zeroing fewer than 0.05% of model weights. Its compact masks also preserve prior-task accuracy better than full fine-tuning, LoRA, and random sparse tuning. (arxiv.org)
Today's best Hacker News stories, summarized and screenshotted, one email a day.