PSI uses an unreleased Anthropic model to solve an 18-year-old quantum conjecture

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, described as the same model shipped at two safeguard levels, MarkTechPost reported. Fable 5.2 scores 49.1% on FrontierCode 2.0, is priced at $8/$40 per million tokens, and comes with a 1M-token context window.

Anthropic's new Fable 5.2 leads the Remote Labor Index from Scale Labs and CAIS at 41.3%, which measures how often AI agents complete real freelance jobs well enough for a paying client to accept. The top score was 2.5% a year ago and about 16% for Fable 5 in July, though product design and animation remain weak spots, The Decoder reports.

Writing in One Useful Thing, Ethan Mollick describes Anthropic's Claude Fable 5.2 as an example of "Jagged AGI": it matches human experts on select tasks such as game design, yet lacks self-directed judgment and taste. His assessment draws on extended hands-on use and finds the model generally expert-level but uneven.

A price study on DEV Community puts Claude Fable 5.2 at $8 per 1M input tokens and $40 per 1M output tokens, 20% below Fable 5.1. Costs are modeled from official Anthropic, OpenAI and Google list prices against a fixed token mix, including a short chat turn, a 400K-token prompt and a cache-heavy coding agent session, rather than measured invoices; benchmark scores are vendor-reported by Anthropic.

Anthropic has released Fable 5.2, its most powerful frontier model, now offered through its API, according to TechPowerUp. The report points to a demo in which the model built a playable first-person shooter from scratch in roughly eight hours, framing it as evidence of how far frontier models can carry a complex project.

Anthropic released two new flagship models, Fable 5.2 and Mythos 5.2, according to Eastmoney. Anthropic's tests show Fable 5.2 reached a 64.9% average success rate on the RoboDojo dual-arm simulated robot benchmark, up from 27.6% for Fable 5.1, versus 39.4% for Claude Opus 5.5 and 29.8% for GPT-6 Astra. Anthropic says the model is substantially better at controlling physical robots, including tasks it was never explicitly trained on.

Anthropic has introduced two new Claude models, Fable 5.2 and Mythos 5.2, which are up to 30% faster than their predecessors, Golem.de reports. The report offers no further detail on parameters, context or availability.

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on Monday, November 9, saying both improve on their 5.1 predecessors in coding, reasoning and vision and generate output 30% faster. Fable 5.2 is generally available, while Mythos 5.2 is limited to organizations approved through Anthropic's Life Sciences and Cyber Verification programmes.

Anthropic has released Claude Fable 5.2 and Claude Mythos 5.2, claiming gains in coding, reasoning, research and cybersecurity, with output up to 30% faster and cost 20% lower. In Anthropic's own benchmarks, Fable 5.2 scores 49.1% on the agentic coding test FrontierCode 2.0, ahead of GPT-6 Astra's 42.8%. The company also says the model coordinated over 1,000 agents to build a bootable operating system from scratch in eight days.

According to Simon Willison, Anthropic's testing shows that when a prompt explicitly claims code ownership, Claude Fable 5.2 tends to actively improve code structure during coding tasks rather than simply repeating existing code.

Anthropic has released Claude Fable 5.1 at the same pricing tier, featuring a 75% reduction in cache read costs and a benchmark score on agentic scientific tasks that more than doubled compared to its predecessor.