gpt.college

Anthropic · Claude 5.5 · Reviewed 2026-10-02

Claude Opus 5.5

Anthropic's Opus update for complex coding and professional workflows, with lower standard token prices.

All models
Overall · 0–10093.5
Status
active
Released
2026-09-22
Context
1,000,000 tokens
Parameters
Undisclosed
Inputs
Text · Image
Type
multimodal
Max output
128,000 tokens
API model ID
claude-opus-5-5
Reasoning
—
Knowledge cutoff
2026-06
Training cutoff
2026-06
Weights / license
Unavailable
Price / 1M tokens
$4 in · $20 out
Cached input
$0.2
Agentic codingComputer use1M-token context
CodingReasoningResearchMultimodalAgentic

Independent benchmark scores

Raw benchmark results

Source-backed rows with the original evaluation context preserved.

18 results
BenchmarkScoreSourceEvaluation notes
AA-Briefcase v1.1Checked 2026-10-02 1822 Anthropic Opus comparison row in Anthropic's Sonnet announcement; AA-Briefcase v1.1 Elo.
AutomationBench-AAChecked 2026-10-02 70% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AA GDP.pdfChecked 2026-10-02 26% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
GDPval-AA v2.1Checked 2026-10-02 1846 Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AA Humanity's Last ExamChecked 2026-10-02 61% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
Artificial Analysis Intelligence Index v4.3.2Checked 2026-10-02 58 Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AA Terminal-Bench 4.0Checked 2026-10-02 60% Artificial Analysis Artificial Analysis Intelligence Index v4.3.2; configuration=max · adaptive reasoning · default fallback. Rounded public score; checked date is not the evaluation execution date.
AutomationBenchChecked 2026-10-02 40% Anthropic Zapier early-access evaluation reported by Anthropic; no fallback models, safeguard interventions counted as failures.
Chartography without ToolsChecked 2026-10-02 64.4% Anthropic Opus comparison row in Anthropic's Sonnet announcement; no tools.
Chartography with ToolsChecked 2026-10-02 89% Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. Tools enabled; separate from no-tools score.
CursorBench 4.0Checked 2026-10-02 57.8% Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. CursorBench 4.0.
FrontierCode 1.1 MainChecked 2026-10-02 54.4% Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. Main set; effort=max.
FrontierCode 1.1 MainChecked 2026-10-02 54.6% Anthropic Anthropic release text; default effort=medium, Main set.
GDPval-AA v2.1Checked 2026-10-02 1846 Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. Artificial Analysis v2.1 Elo, reported by Anthropic.
Humanity's Last Exam with ToolsChecked 2026-10-02 67.7% Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. Tools enabled.
OSWorld 2.1 PartialChecked 2026-10-02 81.8% Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. OSWorld 2.1 partial credit; separate from 2.0.
Terminal-Bench 4.0Checked 2026-10-02 66.4% Anthropic Anthropic table; xhigh effort; standard error ±2.6 points; production safeguards and fallbacks enabled.
Terminal-Bench-Science 0.1Checked 2026-10-02 58.7% Anthropic Anthropic table; adaptive thinking=max, production safeguards and fallbacks enabled. Standard error approximately ±3.5–5 points.

Strengths

  • Agentic coding
  • Computer use
  • 1M-token context

Limitations

  • Adaptive thinking is always enabled
  • Reported benchmark configurations can include safeguard fallbacks; scores depend on effort

Model evidence · official · Checked 2026-10-02

Introducing Claude Opus 5.5

Vendor benchmark table; adaptive thinking at max unless noted, production safeguards and fallbacks enabled except AutomationBench. Preserve benchmark versions.

Model evidence · official · Checked 2026-10-02

Claude Opus 5.5 specifications

Official context, output, API ID, cutoff, default effort and standard pricing; batch beta output limit excluded.

Task lenses preserve raw scores. Radar axes use score / current leader × 100; overall is the equal-weight average of all five indices. Missing evidence is not a zero or a low rank.