Shared benchmarks

26
Shared benchmarks for GPT-5.6 Luna and GPT-6 Luna
BenchmarkGPT-5.6 LunaGPT-6 Luna
  • 43.0%
  • 44.3%
Evaluation details for APEX-Agents Original

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on APEX-Agents Original

GPT-6 Luna

Record for GPT-6 Luna on APEX-Agents Original
  • 1.8%low · OpenAI
  • 4.3%medium · OpenAI
  • 9.1%high · OpenAI
  • 12.9%xhigh · OpenAI
  • 17.0%max · OpenAI
  • 0.2%Zapier AutomationBench Team
  • 1.2%low · OpenAI
  • 9.4%medium · OpenAI
  • 14.5%high · OpenAI
  • 12.6%xhigh · OpenAI
  • 20.7%max · OpenAI
  • 0.3%none · Zapier AutomationBench Team
Evaluation details for AutomationBench 1.0.6

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on AutomationBench 1.0.6
Record for GPT-5.6 Luna on AutomationBench 1.0.6
Record for GPT-5.6 Luna on AutomationBench 1.0.6
Record for GPT-5.6 Luna on AutomationBench 1.0.6
Record for GPT-5.6 Luna on AutomationBench 1.0.6
Record for GPT-5.6 Luna on AutomationBench 1.0.6

GPT-6 Luna

Record for GPT-6 Luna on AutomationBench 1.0.6
Record for GPT-6 Luna on AutomationBench 1.0.6
Record for GPT-6 Luna on AutomationBench 1.0.6
Record for GPT-6 Luna on AutomationBench 1.0.6
Record for GPT-6 Luna on AutomationBench 1.0.6
Record for GPT-6 Luna on AutomationBench 1.0.6
  • 83.8%
  • 83.3%
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.

GPT-5.6 Luna

GPT-6 Luna

  • 31.4%
  • 29.1%
Evaluation details for Chartography October 2026 leaderboard

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Chartography October 2026 leaderboard

GPT-6 Luna

Record for GPT-6 Luna on Chartography October 2026 leaderboard
  • 92.9%
  • 90.9%
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.
  • 67.0%max · 1.1 · DataCurve
  • 1.2%low · 1.1 · OpenAI
  • 9.3%medium · 1.1 · OpenAI
  • 42.4%high · 1.1 · OpenAI
  • 56.2%xhigh · 1.1 · OpenAI
  • 62.2%max · 1.1 · OpenAI
  • 67.2%1.1 · OpenAI
  • 2.4%low · 1.1 · OpenAI
  • 44.5%medium · 1.1 · OpenAI
  • 59.3%high · 1.1 · OpenAI
  • 61.3%xhigh · 1.1 · OpenAI
  • 66.6%max · 1.1 · OpenAI
Potentially non-equivalent · Evaluation details for DeepSWE 1.1

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on DeepSWE 1.1
Record for GPT-5.6 Luna on DeepSWE 1.1
Record for GPT-5.6 Luna on DeepSWE 1.1
Record for GPT-5.6 Luna on DeepSWE 1.1
Record for GPT-5.6 Luna on DeepSWE 1.1
Record for GPT-5.6 Luna on DeepSWE 1.1
Record for GPT-5.6 Luna on DeepSWE 1.1

GPT-6 Luna

Record for GPT-6 Luna on DeepSWE 1.1
Record for GPT-6 Luna on DeepSWE 1.1
Record for GPT-6 Luna on DeepSWE 1.1
Record for GPT-6 Luna on DeepSWE 1.1
Record for GPT-6 Luna on DeepSWE 1.1
  • 64.3%
  • 67.0%
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.

GPT-5.6 Luna

Record for GPT-5.6 Luna on DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps

GPT-6 Luna

Record for GPT-6 Luna on DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps
  • 55.0%
  • 49.9%
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.

  • 63.6%none
  • 82.3%low
  • 91.6%max
  • 90.5%
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.

GPT-5.6 Luna

Record for GPT-5.6 Luna on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness

GPT-6 Luna

Record for GPT-6 Luna on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
  • 91.8%
  • 89.9%
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.

GPT-5.6 Luna

Record for GPT-5.6 Luna on Graduate-Level Google-Proof Q&A Diamond — Mercor single-shot

GPT-6 Luna

Record for GPT-6 Luna on Graduate-Level Google-Proof Q&A Diamond — Mercor single-shot
  • 1.3%
  • 2.9%
Evaluation details for HLAB October 2026 evaluation

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Harvey's Legal Agent Benchmark October 2026 evaluation

GPT-6 Luna

Record for GPT-6 Luna on Harvey's Legal Agent Benchmark October 2026 evaluation
  • 39.6%
  • 38.7%
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.

GPT-5.6 Luna

GPT-6 Luna

  • 70.2%
  • 64.8%
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.

GPT-5.6 Luna

Record for GPT-5.6 Luna on MedXpertQA MM — Mercor single-shot

GPT-6 Luna

Record for GPT-6 Luna on MedXpertQA MM — Mercor single-shot
  • 86.5%
  • 86.2%
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.

GPT-5.6 Luna

GPT-6 Luna

  • 79.1%
  • 76.6%
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.

GPT-5.6 Luna

Record for GPT-5.6 Luna on MMMU-Pro 1,730 questions — Mercor single-shot

GPT-6 Luna

Record for GPT-6 Luna on MMMU-Pro 1,730 questions — Mercor single-shot
  • 12.0%low
  • 22.7%medium
  • 35.9%high
  • 48.0%xhigh
  • 52.7%max
  • 8.3%low
  • 31.5%medium
  • 41.4%high
  • 46.7%xhigh
  • 52.7%max
Evaluation details for OSWorld 2.0 v2026.08.08 Offline — Partial

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

GPT-5.6 Luna

GPT-6 Luna

Record for GPT-6 Luna on OSWorld 2.0 v2026.08.08 Offline — Partial
Record for GPT-6 Luna on OSWorld 2.0 v2026.08.08 Offline — Partial
  • 41.0%
  • 41.4%
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.

  • 46.0%
  • 33.9%
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.

  • 92.4%
  • 78.2%
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.

  • 76.2%
  • 76.1%
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.

GPT-5.6 Luna

GPT-6 Luna

  • 79.0%
  • 73.0%
Evaluation details for Terminal-Bench 2.1 — Vals AI Terminus 2

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Terminal-Bench 2.1 — Vals AI Terminus 2

GPT-6 Luna

Record for GPT-6 Luna on Terminal-Bench 2.1 — Vals AI Terminus 2
  • 17.3%
  • 16.4%
Evaluation details for Terminal-Bench 4.0 — Codex

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Terminal-Bench 4.0 — Codex

GPT-6 Luna

Record for GPT-6 Luna on Terminal-Bench 4.0 — Codex
  • 10.1%
  • 13.5%
Evaluation details for Terminal-Bench 4.0 — Mercor mini-swe-agent

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Terminal-Bench 4.0 — Mercor mini-swe-agent

GPT-6 Luna

Record for GPT-6 Luna on Terminal-Bench 4.0 — Mercor mini-swe-agent
  • 11.6%
  • 13.6%
Evaluation details for Terminal-Bench 4.0 — Vals AI mini-swe-agent

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Terminal-Bench 4.0 — Vals AI mini-swe-agent

GPT-6 Luna

Record for GPT-6 Luna on Terminal-Bench 4.0 — Vals AI mini-swe-agent
  • 51.7%
  • 51.2%
Evaluation details for Vals Index 2.1

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

GPT-5.6 Luna

Record for GPT-5.6 Luna on Vals Index 2.1

GPT-6 Luna

Record for GPT-6 Luna on Vals Index 2.1
  • 77.1%
  • 81.6%
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.