Free guide · Sep 22, 2026
Opus 5.5: what the benchmarks really show
Anthropic's launch chart read honestly: where Opus 5.5 is better, and where it isn't.
The short version
Anthropic released Claude Opus 5.5 on September 22, 2026. On Anthropic's own launch table it scores above Fable 5.1 on all nine benchmarks in that table, and it costs \$4 per million input tokens and \$20 per million output. Fable 5.1 is \$10 and \$50.
Anthropic adds one caveat right above that table: in their own use, the gap between Opus 5.5 and Fable 5.1 is narrower than the scores suggest. So I read it as better on paper, close in practice, and a lot cheaper. That's enough for me to make it my default and keep Fable 5.1 for the jobs where I've already seen it earn its price. Claude Code already treats it as the default Opus model.
Prices from Anthropic's pricing docs, checked September 22, 2026.
What's in the full guide
- 011. Switch Claude Code to Opus 5.5
- 022. The settings that keep it cheap
- 033. When Fable 5.1 is still worth it
- 044. Run your own side by side in fifteen minutes
- 055. If you're on a Claude plan
- 06Sources