[AINews] Claude Fable/Mythos 5.1: new SOTA model, 75% cache price cut but 70% more output tokens
Anthropic’s 5.1 launch looks stronger on frontier benchmarks, but the economics are mixed once longer outputs are counted.
Latent Space says Claude Fable 5.1 and Mythos 5.1 kept headline token prices unchanged while cutting cache reads 75%, a meaningful break for long-context agent work. Artificial Analysis put Fable 5.1 at the top of its Intelligence Index, but also found it used about 1.7x more output tokens, making per-task cost about 20% higher than Fable 5. The note flags caveats around server-side fallback, safety routing, and whether Fable and Mythos are distinct models or policy paths around the same weights. User reaction split between strong praise for coding and planning gains and complaints about rate limits, safeguards, and subscription UX. Latent Space's note
Latent Space says Claude Fable 5.1 and Mythos 5.1 kept headline token prices unchanged while cutting cache reads 75%, a meaningful break for long-context agent work. Artificial Analysis put Fable 5.1 at the top of its Intelligence Index, but also found it used about 1.7x more output tokens, making per-task cost about 20% higher than Fable 5. The note flags caveats around server-side fallback, safety routing, and whether Fable and Mythos are distinct models or policy paths around the same weights. User reaction split between strong praise for coding and planning gains and complaints about rate limits, safeguards, and subscription UX. Latent Space's note
score 8