An
claude-fable-5-1
Benchmark & Insights
Anthropic Claude API
Updated Sep 27, 2026 All models
Sample size
58 runs
in window
Accuracy
84.5%
consensus match · 58d
Confidence
85%
over 58 runs
Window end
Sep 27, 2026
most recent run
Input price
$10.00/MTok
prompt tokens
Output price
$50.00/MTok
completion tokens
Model insights
- 01 A clear step back from claude-fable-5 over the exact same 58 days, with many more misses, all in the optimistic direction.
- 02 It keeps saying "low" on "medium" days and said "safe" on two "unsafe" days, which is a poor result at the top price tier; the short window still makes this a provisional read.
Notes
Comparing to the previous version - there are breaking changes for Claude Fable 5.1
Recent forecasts