Anthropic releases Fable 5.2 and Mythos 5.2, claiming breakthroughs in black hole physics

Bulletin

Simon Willison’s weblog reported on November 9 the release of llm-anthropic 0.31, the plugin update marking Anthropic’s launch of two new models, Fable 5.2 and Mythos 5.2. According to Willison’s post, Fable 5.2 is currently ranked first on the Artificial Analysis independent benchmarking site.

Willison’s summary notes that Anthropic’s own announcement describes the release as having “solved two open problems in the physics of black holes.” Willison, in his post, responded to this framing with visible skepticism, suggesting the claim says more about Anthropic’s internal benchmarking standards than about settled physics.

No further detail from Anthropic’s announcement is available in the material reviewed here. The claim’s substance, method, or peer review status is not described in the source. Willison’s post functions primarily as a changelog entry for the llm-anthropic command line tool, noting the new model identifiers now supported by the plugin, with the benchmark ranking and the black hole claim mentioned as context for the release rather than independently verified.

Readers are referred to Anthropic’s own release page and to the Artificial Analysis leaderboard, both linked in Willison’s original post, for primary claims. This log records only what Willison’s bulletin states and does not extend beyond it.