Claude Opus 5.5 Ranks High on Intelligence but Comes at a Premium Price

A benchmark analysis of Anthropic's Claude Opus 5.5, running in its Adaptive Reasoning, Max Effort, Default Fallback configuration, has been published comparing the model against competitors on quality, price, performance and context window.

According to the analysis, Claude Opus 5.5 supports both text and image input, produces text output, and offers a context window of 1 million tokens. The model scored 58 on the Artificial Analysis Intelligence Index, well above the median score of 25 among comparable models, positioning it among the leading models for intelligence.

The report notes that Claude Opus 5.5 is notably verbose. During evaluation on the Intelligence Index, it generated 260 million tokens, roughly three times the median of 88 million tokens generated by other models tested.

That verbosity, combined with the model's per-token rates, contributes to a higher overall cost. Claude Opus 5.5 is priced at $4.00 per million input tokens, compared with a median of $2.00, and $20.00 per million output tokens, against a median of $10.00. Both rates are described in the analysis as "somewhat expensive" relative to similarly positioned models.

The combination of high token output and elevated per-token pricing meant that running the full Intelligence Index evaluation on Claude Opus 5.5 cost $8,708.20 in total, according to the report.

The analysis, as reported via Hacker News, credits Artificial Analysis as the source of the benchmark data. Some details of the underlying report were not fully extractable from the original material, and specific methodology, additional comparison models, and performance metrics such as tokens per second and time to first token were referenced but not fully detailed in the available text. Readers seeking the complete breakdown should consult the original Artificial Analysis report directly.