PyyanThe AI Index

Pyyan / Compare / Claude Opus 5 vs Claude Opus 5.5

Claude Opus 5 vs Claude Opus 5.5

2 of 5

Language Models · verified 24 Sept 2026

Claude Opus 5Anthropiccurrent
Claude Opus 5.5Anthropiccurrent
3 slots left
SpecificationClaude Opus 5Claude Opus 5.5
SummaryComplex agentic coding and enterprise work at half the cost of Fable 5.Fable 5.1's level of work on most tasks, at 40% less to run than Opus 5.
Context1M1M
Input ($/Mtok)$5$4
Output ($/Mtok)$25$20
LicenceProprietaryProprietary
Max output128K128K
Parameters—Not published
Released24 Jul 202622 Sep 2026
Model IDclaude-opus-5claude-opus-5-5
CategoryLanguage ModelsLanguage Models
OfficialAnthropic ↗Anthropic ↗

Highlighted rows are where these differ.

Claude Opus 5

  • Thinking on by default — a step change over Opus 4.8
  • 512-token prompt-cache minimum, down from 1024
  • Separate rate-limit bucket from the Opus 4.x pool

Best for agentic coding.

Full spec sheet →

Claude Opus 5.5

  • 66.4% on Terminal-Bench 4.0 against 52.3% for Opus 5, on Anthropic's own harness
  • Cache reads cut 60% to $0.20, and output more than 30% faster than Opus 5
  • Attempts to work around its boundaries 85% less often than Opus 5, Anthropic reports

Best for agentic coding at flagship quality.

Full spec sheet →