Shared benchmarks

36
Shared benchmarks for Claude Opus 4.8 and Claude Opus 5
BenchmarkClaude Opus 4.8Claude Opus 5
  • 48.9%
  • 65.8%
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 4.8

Record for Claude Opus 4.8 on APEX-Agents Original

Claude Opus 5

Record for Claude Opus 5 on APEX-Agents Original
  • 13.6%low
  • 14.0%medium
  • 16.1%high
  • 16.9%xhigh
  • 16.4%max
  • 20.4%low
  • 23.9%medium
  • 20.6%high
  • 25.3%xhigh
  • 26.9%max
Evaluation details for AutomationBench 1.0.6

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

Claude Opus 4.8

Record for Claude Opus 4.8 on AutomationBench 1.0.6
Record for Claude Opus 4.8 on AutomationBench 1.0.6
Record for Claude Opus 4.8 on AutomationBench 1.0.6
Record for Claude Opus 4.8 on AutomationBench 1.0.6
Record for Claude Opus 4.8 on AutomationBench 1.0.6

Claude Opus 5

Record for Claude Opus 5 on AutomationBench 1.0.6
Record for Claude Opus 5 on AutomationBench 1.0.6
Record for Claude Opus 5 on AutomationBench 1.0.6
Record for Claude Opus 5 on AutomationBench 1.0.6
Record for Claude Opus 5 on AutomationBench 1.0.6
  • 73.7%
  • 84.6%
Evaluation details for BrowseComp 130-question subset — Mercor web research agent

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

Claude Opus 5

  • 15.9%
  • 27.3%
Evaluation details for Chartography October 2026 leaderboard

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

Claude Opus 4.8

Record for Claude Opus 4.8 on Chartography October 2026 leaderboard

Claude Opus 5

Record for Claude Opus 5 on Chartography October 2026 leaderboard
  • 93.2%
  • 94.6%
Evaluation details for CharXiv Descriptive and reasoning — Mercor single-shot

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

DeepSWE 1.1Potentially non-equivalent: Evaluator sets differ.
  • 59.0%1.1 · DataCurve
  • 57.7%low · 1.1 · Anthropic
  • 66.9%medium · 1.1 · Anthropic
  • 68.0%high · 1.1 · Anthropic
  • 69.7%xhigh · 1.1 · Anthropic
  • 68.8%max · 1.1 · Anthropic
  • 74.0%max · 1.1 · DataCurve
Potentially non-equivalent · Evaluation details for DeepSWE 1.1

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

Claude Opus 4.8

Record for Claude Opus 4.8 on DeepSWE 1.1

Claude Opus 5

Record for Claude Opus 5 on DeepSWE 1.1
Record for Claude Opus 5 on DeepSWE 1.1
Record for Claude Opus 5 on DeepSWE 1.1
Record for Claude Opus 5 on DeepSWE 1.1
Record for Claude Opus 5 on DeepSWE 1.1
Record for Claude Opus 5 on DeepSWE 1.1
  • 59.6%
  • 70.2%xhigh
  • 71.4%max
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 4.8

Claude Opus 5

Record for Claude Opus 5 on DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps
Record for Claude Opus 5 on DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps
  • 53.9%
  • 58.6%
Evaluation details for Finance Agent Benchmark 2 — October 2026 evaluation

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

  • 85.4%none
  • 88.4%low
  • 91.0%max
  • 87.9%low
  • 93.9%max
  • 92.9%
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 4.8

Claude Opus 5

Record for Claude Opus 5 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
  • 92.0%
  • 93.6%
Evaluation details for GPQA Diamond — Mercor single-shot

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

Claude Opus 4.8

Claude Opus 5

Record for Claude Opus 5 on Graduate-Level Google-Proof Q&A Diamond — Mercor single-shot
  • 92.4%
  • 93.4%
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 4.8

Record for Claude Opus 4.8 on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness

Claude Opus 5

Record for Claude Opus 5 on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness
  • 9.6%
  • 6.7%
Evaluation details for HLAB October 2026 evaluation

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

Claude Opus 4.8

Record for Claude Opus 4.8 on Harvey's Legal Agent Benchmark October 2026 evaluation

Claude Opus 5

Record for Claude Opus 5 on Harvey's Legal Agent Benchmark October 2026 evaluation
  • 49.8%
  • 47.8%low
  • 54.2%medium
  • 56.0%high
  • 56.4%xhigh
  • 56.3%max
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 4.8

Record for Claude Opus 4.8 on Humanity's Last Exam Full set — No tools

Claude Opus 5

Record for Claude Opus 5 on Humanity's Last Exam Full set — No tools
  • 57.9%
  • 56.1%low
  • 61.3%medium
  • 63.2%high
  • 64.8%xhigh
  • 64.7%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 4.8

Record for Claude Opus 4.8 on Humanity's Last Exam Full set — With tools

Claude Opus 5

  • 44.8%high
  • 37.1%max
  • 52.0%high
  • 53.2%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.

  • 82.2%
  • 85.8%
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 4.8

Claude Opus 5

Record for Claude Opus 5 on MCP Atlas Public, April 2026 update
  • 72.5%
  • 79.8%
Evaluation details for MedXpertQA MM — Mercor single-shot

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

Claude Opus 4.8

Record for Claude Opus 4.8 on MedXpertQA MM — Mercor single-shot

Claude Opus 5

Record for Claude Opus 5 on MedXpertQA MM — Mercor single-shot
  • 90.0%
  • 92.0%
Evaluation details for MMLU-Pro Original — Mercor single-shot

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

Claude Opus 4.8

Claude Opus 5

  • 89.6%
  • 91.6%
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 4.8

Record for Claude Opus 4.8 on MMLU-Pro Original — Vals AI harness

Claude Opus 5

Record for Claude Opus 5 on MMLU-Pro Original — Vals AI harness
  • 79.5%
  • 84.2%
Evaluation details for MMMU-Pro 1,730 questions — Mercor single-shot

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

Claude Opus 4.8

Record for Claude Opus 4.8 on MMMU-Pro 1,730 questions — Mercor single-shot

Claude Opus 5

Record for Claude Opus 5 on MMMU-Pro 1,730 questions — Mercor single-shot
  • 86.6%
  • 89.9%
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 4.8

Record for Claude Opus 4.8 on MMMU-Pro Vals AI harness

Claude Opus 5

Record for Claude Opus 5 on MMMU-Pro Vals AI harness
  • 33.8%High
  • 32.9%Max
  • 35.0%
Evaluation details for PostTrainBench 1.1 — Claude Code

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

Claude Opus 4.8

Record for Claude Opus 4.8 on PostTrainBench 1.1 — Claude Code
Record for Claude Opus 4.8 on PostTrainBench 1.1 — Claude Code

Claude Opus 5

Record for Claude Opus 5 on PostTrainBench 1.1 — Claude Code
  • 30.7%High
  • 31.0%Max
  • 35.7%
Evaluation details for PostTrainBench 1.2 — Claude Code

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

Claude Opus 4.8

Record for Claude Opus 4.8 on PostTrainBench 1.2 — Claude Code
Record for Claude Opus 4.8 on PostTrainBench 1.2 — Claude Code

Claude Opus 5

Record for Claude Opus 5 on PostTrainBench 1.2 — Claude Code
  • 47.2%
  • 68.0%
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 4.8

Record for Claude Opus 4.8 on Riemann-bench October 2026 leaderboard

Claude Opus 5

Record for Claude Opus 5 on Riemann-bench October 2026 leaderboard
  • 53.0%
  • 59.9%
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.

  • 44.4%
  • 50.3%
Evaluation details for SWE Atlas Codebase QnA — Mercor mini-swe-agent, 250 steps

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

  • 57.3%
  • 63.2%
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.

  • 84.4%
  • 89.5%
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 4.8

Record for Claude Opus 4.8 on SWE-bench Multilingual

Claude Opus 5

Record for Claude Opus 5 on SWE-bench Multilingual
  • 86.6%
  • 96.2%
Evaluation details for SWE-bench Multilingual — Mercor mini-swe-agent, 1000 steps

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

  • 88.6%
  • 96.0%
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 4.8

Record for Claude Opus 4.8 on SWE-bench Verified

Claude Opus 5

Record for Claude Opus 5 on SWE-bench Verified
  • 88.5%
  • 93.5%
Evaluation details for SWE-bench Verified — Mercor mini-swe-agent, 1000 steps

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

Claude Opus 5

  • 88.6%
  • 97.0%
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 4.8

Claude Opus 5

Record for Claude Opus 5 on SWE-bench Verified — Vals AI mini-swe-agent
  • 23.6%
  • 34.9%low
  • 44.9%medium
  • 50.3%high
  • 53.9%xhigh
  • 51.8%max
Evaluation details for Terminal-Bench 4.0 — Claude Code

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

Claude Opus 4.8

Record for Claude Opus 4.8 on Terminal-Bench 4.0 — Claude Code

Claude Opus 5

Record for Claude Opus 5 on Terminal-Bench 4.0 — Claude Code
Record for Claude Opus 5 on Terminal-Bench 4.0 — Claude Code
Record for Claude Opus 5 on Terminal-Bench 4.0 — Claude Code
Record for Claude Opus 5 on Terminal-Bench 4.0 — Claude Code
Record for Claude Opus 5 on Terminal-Bench 4.0 — Claude Code
  • 55.1%
  • 63.7%
Evaluation details for Vals Index 2.1

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

Claude Opus 4.8

Record for Claude Opus 4.8 on Vals Index 2.1

Claude Opus 5

Record for Claude Opus 5 on Vals Index 2.1
  • 82.7%
  • 88.4%
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.

  • 39.7%
  • 48.7%
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.