Lokendra Singh.
← ALL WRITINGS
// Writing · Sep 23, 2026 · 2 min read

Claude Opus 5.5 at 40% of Fable's price

claudellms

Claude Opus 5.5 launched on 22 September. It beat Fable 5.1 on every benchmark Anthropic published, at 40% of Fable's token price.

Before anyone asks: there is no Opus 5.1. It went Opus 5 (24 July) → Opus 5.5.

Price per 1M tokens

Model Input Output Cache read
Opus 5.5 $4 $20 $0.20
Opus 5 $5 $25 $0.50
Fable 5.1 $10 $50 $0.25

Anthropic says ~40% cheaper per task than Opus 5, and 30%+ faster output.

Opus 5.5 vs Fable 5.1 vs Opus 5

Benchmark Opus 5.5 Fable 5.1 Opus 5
Terminal-Bench 4.0 66.4% 55.8% 52.3%
CursorBench 4.0 57.8% 51.8% 46.6%
GDPval-AA (real work) 1846 1735 1708
OSWorld 2.0 (computer use) 81.8% 80.7% 74.0%

What that looks like in practice

  • Early tester: a 200K-line codebase audited and fixed in under 3 hours. Opus 5: 20+ hours.
  • Anthropic's own test: HAProxy ported from C to Rust in 9.5 hours vs Fable 5.1's 12, at 51% lower cost.
  • Deloitte: at its lowest effort it caught 72% of known bugs, vs 56% for Opus 5 at high effort.

What else changed

  • First Opus with Fable-class safeguards. Most cyber requests fall back to Opus 4.8.
  • Thinking can no longer be switched off.
  • Sonnet 5.5 and Haiku 5.5 "in the coming weeks".

The honest part

  • Every score above is Anthropic-reported. Wait for independent leaderboards before trusting the margins.
  • Anthropic itself says the real-world gap to Fable 5.1 is narrower than these numbers suggest.
  • Anthropic also flags that the model often suspects it is being evaluated.

My take

Opus 5.5 is the new default. Fable 5.1 just became a specialist tool you reach for only after Opus 5.5 fails on your task.

Is Opus 5.5 replacing Fable or Opus 5 in your stack?

Sources