DATUM

Anthropic releases Claude Fable 5.1 with five reasoning effort levels

single source · 1 articles · capability · confidence: high · first seen 2026-09-01 23:57 UTC

Anthropic has released Claude Fable 5.1, the latest model in its Fable line. The company says it sets a new standard for coding, knowledge work and long-running problem-solving. On the Terminal-Bench-Science 0.1 benchmark, announced 27 August, it scores 52.6%, up from 24.7% for Fable 5, 29.0% for Opus 5 and 22.4% for GPT-5.6 Sol. Other benchmarks show slightly improved scores. The model has five reasoning effort levels—low, medium, high, xhigh and max—with no option to disable reasoning entirely. Simon Willison's testing found the low and medium settings produced no visible reasoning tokens for a specific SVG prompt.

What this means for you

If you build on Anthropic's API, test Fable 5.1's low and medium effort levels before assuming you need high. The source found those two produced no reasoning traces and slightly lower cost for an SVG prompt, though results may vary.

Key facts

  • ·Anthropic says Claude Fable 5.1 scores 52.6% on the Terminal-Bench-Science 0.1 benchmark. source
  • ·Previous scores on that benchmark: Fable 5 24.7%, Opus 5 29.0%, GPT-5.6 Sol 22.4%. source
  • ·Claude Fable 5.1 has five reasoning levels: low, medium, high, xhigh and max, with no option to turn off reasoning entirely. source
  • ·The Terminal-Bench-Science 0.1 benchmark was first announced on 27 August. source
  • ·In Simon Willison's test, low and medium settings produced no visible reasoning tokens, with output token counts of 1,998 and 1,977 respectively. source

What the sources say

  • Simon Willison — Tests Fable 5.1's reasoning levels with SVG generation and reports low/medium produced no reasoning traces for that prompt.

Sources

The original reporting. Follow these — they did the work.

← the wire