Shared benchmarks

25
Shared benchmarks for Claude Opus 4.6 and Claude Opus 4.7
BenchmarkClaude Opus 4.6Claude Opus 4.7
  • 46.3%
  • 49.2%
Evaluation details for APEX-Agents Original

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on APEX-Agents Original

Claude Opus 4.7

Record for Claude Opus 4.7 on APEX-Agents Original
  • 31.9%
  • 49.3%
Evaluation details for DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Claude Opus 4.7

  • 91.3%
  • 94.2%
Evaluation details for GPQA Diamond

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Graduate-Level Google-Proof Q&A Diamond

Claude Opus 4.7

Record for Claude Opus 4.7 on Graduate-Level Google-Proof Q&A Diamond
  • 88.4%
  • 90.2%xhigh
  • 86.4%max
Evaluation details for GPQA Diamond — Epoch AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Claude Opus 4.7

  • 89.6%
  • 90.2%
Evaluation details for GPQA Diamond — Vals AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness

Claude Opus 4.7

Record for Claude Opus 4.7 on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness
  • 40.0%
  • 46.9%
Evaluation details for Humanity's Last Exam Full set — No tools

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Humanity's Last Exam Full set — No tools

Claude Opus 4.7

Record for Claude Opus 4.7 on Humanity's Last Exam Full set — No tools
  • 19.0%Non-Thinking
  • 34.4%thinking-max
  • 36.2%
Evaluation details for Humanity's Last Exam Full set — Scale AI evaluation

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

  • 53.0%
  • 43.0%adaptive thinking, low
  • 48.4%adaptive thinking, medium
  • 53.2%adaptive thinking, high
  • 55.4%adaptive thinking, xhigh
  • 54.7%adaptive thinking, max
Evaluation details for Humanity's Last Exam Full set — With tools

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Humanity's Last Exam Full set — With tools

Claude Opus 4.7

Record for Claude Opus 4.7 on Humanity's Last Exam Full set — With tools
Record for Claude Opus 4.7 on Humanity's Last Exam Full set — With tools
Record for Claude Opus 4.7 on Humanity's Last Exam Full set — With tools
Record for Claude Opus 4.7 on Humanity's Last Exam Full set — With tools
Record for Claude Opus 4.7 on Humanity's Last Exam Full set — With tools
  • 35.0%
  • 42.2%high
  • 35.4%max
Evaluation details for Humanity's Last Exam Text only — Mercor single-shot

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

  • 76.8%
  • 79.1%
Evaluation details for MCP Atlas Public, April 2026 update

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Claude Opus 4.7

  • 89.1%
  • 89.9%
Evaluation details for MMLU-Pro Original — Vals AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Claude Opus 4.7

Record for Claude Opus 4.7 on MMLU-Pro Original — Vals AI harness
  • 83.9%
  • 85.5%
Evaluation details for MMMU-Pro Vals AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on MMMU-Pro Vals AI harness

Claude Opus 4.7

Record for Claude Opus 4.7 on MMMU-Pro Vals AI harness
  • 27.2%
  • 32.8%
Evaluation details for Riemann-bench October 2026 leaderboard

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Riemann-bench October 2026 leaderboard

Claude Opus 4.7

Record for Claude Opus 4.7 on Riemann-bench October 2026 leaderboard
  • 47.0%
  • 51.7%
Evaluation details for SimpleQA Verified Original — Epoch AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

  • 33.3%
  • 40.3%
Evaluation details for SWE Atlas Codebase QnA — Public; Claude Code

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Claude Opus 4.7

  • 35.6%
  • 48.6%
Evaluation details for SWE Atlas Refactoring — Claude Code

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on SWE Atlas Refactoring — Claude Code

Claude Opus 4.7

Record for Claude Opus 4.7 on SWE Atlas Refactoring — Claude Code
SWE AtlasPotentially non-equivalent: Benchmark versions differ. Evaluator sets differ.
  • 30.0%Codebase QnA — Public; mini-swe-agent · Scale AI
  • 47.0%Codebase QnA — Mercor mini-swe-agent, 250 steps · Mercor
Potentially non-equivalent · Evaluation details for SWE Atlas

Benchmark versions differ. Evaluator sets differ. Evaluation methodology is not recorded by the registry.

  • 77.8%
  • 80.5%
Evaluation details for SWE-bench Multilingual

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on SWE-bench Multilingual

Claude Opus 4.7

Record for Claude Opus 4.7 on SWE-bench Multilingual
  • 80.8%
  • 87.6%
Evaluation details for SWE-bench Verified

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on SWE-bench Verified

Claude Opus 4.7

Record for Claude Opus 4.7 on SWE-bench Verified
  • 75.6%
  • 83.5%
Evaluation details for SWE-bench Verified — Epoch AI harness

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on SWE-bench Verified — Epoch AI harness

Claude Opus 4.7

  • 78.2%
  • 82.0%
Evaluation details for SWE-bench Verified — Vals AI mini-swe-agent

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Claude Opus 4.7

  • 65.4%
  • 69.4%
Evaluation details for Terminal-Bench 2.0

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Terminal-Bench 2.0

Claude Opus 4.7

Record for Claude Opus 4.7 on Terminal-Bench 2.0
  • 58.4%
  • 68.5%
Evaluation details for Terminal-Bench 2.0 — Vals AI Terminus 2

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

Claude Opus 4.6

Record for Claude Opus 4.6 on Terminal-Bench 2.0 — Vals AI Terminus 2

Claude Opus 4.7

Record for Claude Opus 4.7 on Terminal-Bench 2.0 — Vals AI Terminus 2
  • 57.6%
  • 53.5%thinking
  • 71.0%
Evaluation details for Vibe Code Bench 1.1 — OpenHands, October 2026 evaluation

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.

  • 27.3%
  • 40.2%
Evaluation details for τ³-bench τ³ — Banking, AllTools retrieval, v1.0.1 grading

The recorded benchmark version, metric, and evaluator sets match. Evaluation methodology is not recorded by the registry.