Fable is cheaper than 5.1. Not cheaper than Opus. Anthropic shipped Claude Fable
By Jaspinall · November 10, 2026 · Curated by George's Blog
Fable is cheaper than 5.1.
Not cheaper than Opus.
Anthropic shipped Claude Fable 5.2 and Claude Mythos 5.2 today.
Fable 5.2 is $8 per million input tokens and $40 per million output. That's 20% under Fable 5.1's $10 and $50.
Opus 5.5 is still $4 and $20. So Fable costs twice what Anthropic calls its strongest daily driver.
Anthropic's own advice matches that math. Start with Opus 5.5 for everyday coding and knowledge work. Move up to Fable when the job gets hard and open-ended. Move down to Sonnet or Haiku when it doesn't.
Fable and Mythos are the same underlying model. Fable adds cyber and biology safeguards so it can ship on paid Claude plans, Claude Code, the API and the major clouds. Mythos stays with vetted cyber and life-sciences orgs, with Claude Security as the Enterprise route.
"Up to 30% faster" and fast mode's "up to 2.5x" are about output speed, not how long a whole task takes. Fast mode costs $16/$80, isn't listed on Bedrock, and can't use Batch.
The chart below is Anthropic's FrontierCode export from the announcement page. At High effort, Fable 5.2 hits 45.3% for about $6.72 per task. Opus 5.5's best is 45.5% for about $12.82. Cognition ran both Claude models in Claude Code. Competitor models used different agent software, so this is not a blanket pure-model comparison. These are launch-day, vendor-published results. I haven't seen independent replication yet. At max effort, timeouts change the score, so higher effort is not always better.
A 20% cut from the previous Fable price does not make Fable your new default. Keep Opus on the daily work. Put Fable on the hard technical jobs where the extra spend has a chance to earn it back. Leave Sonnet or Haiku on the bulk work, and only pull Fable in as an advisor when the hard step is worth the tokens.
Test that on one of your own workflows before you change the default model in production.