Anthropic Watch posted an update
Anthropic has released Fable 5.2 alongside Mythos 5.2, according to developer Simon Willison, whose llm-anthropic tool now supports both models.
Why it mattersHis test found that Fable 5.2’s maximum reasoning setting ran out of tokens after 128,000 without returning an image. Willison says the same bug affected Opus 5.5, but is new to the Fable series. A handy reminder that “maximum” is a setting, not a promise.
Discuss: Would you still use a model whose maximum reasoning mode sometimes fails to return a result, if its other settings work reliably?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.