Shared benchmarks

21
Shared benchmarks for GPT-5.6 Sol and GPT-6 Sol
BenchmarkGPT-5.6 SolGPT-6 Sol
  • 11.7%low · OpenAI
  • 19.6%medium · OpenAI
  • 24.8%high · OpenAI
  • 26.3%xhigh · OpenAI
  • 28.8%max · OpenAI
  • 3.8%Zapier AutomationBench Team
  • 21.2%low · OpenAI
  • 26.9%medium · OpenAI
  • 31.2%high · OpenAI
  • 33.2%xhigh · OpenAI
  • 32.0%max · OpenAI
  • 9.1%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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on AutomationBench 1.0.6
Record for GPT-6 Sol on AutomationBench 1.0.6
Record for GPT-6 Sol on AutomationBench 1.0.6
Record for GPT-6 Sol on AutomationBench 1.0.6
Record for GPT-6 Sol on AutomationBench 1.0.6
Record for GPT-6 Sol on AutomationBench 1.0.6
  • 90.6%
  • 89.4%
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 Sol

GPT-6 Sol

  • 45.0%
  • 53.6%
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 Sol

Record for GPT-5.6 Sol on Chartography October 2026 leaderboard

GPT-6 Sol

Record for GPT-6 Sol on Chartography October 2026 leaderboard
DeepSWE 1.1Potentially non-equivalent: Evaluator sets differ.
  • 73.0%max · 1.1 · DataCurve
  • 45.4%low · 1.1 · OpenAI
  • 61.1%medium · 1.1 · OpenAI
  • 69.4%high · 1.1 · OpenAI
  • 70.7%xhigh · 1.1 · OpenAI
  • 72.7%1.1 · OpenAI
  • 37.2%low · 1.1 · OpenAI
  • 56.6%medium · 1.1 · OpenAI
  • 65.3%high · 1.1 · OpenAI
  • 66.6%xhigh · 1.1 · OpenAI
  • 68.8%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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on DeepSWE 1.1
Record for GPT-6 Sol on DeepSWE 1.1
Record for GPT-6 Sol on DeepSWE 1.1
Record for GPT-6 Sol on DeepSWE 1.1
Record for GPT-6 Sol on DeepSWE 1.1
  • 70.5%
  • 70.5%
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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on DeepSWE 1.1 — Mercor mini-swe-agent, 500 steps
  • 53.8%
  • 49.1%
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.

  • 82.8%none
  • 89.9%low
  • 93.5%max
  • 94.3%
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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
  • 2.5%
  • 1.7%
Evaluation details for HLAB October 2026 evaluation

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

GPT-5.6 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on Harvey's Legal Agent Benchmark October 2026 evaluation
  • 47.0%
  • 49.2%
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 Sol

GPT-6 Sol

MMLU-ProPotentially non-equivalent: Benchmark versions differ. Evaluator sets differ.
  • 89.1%Original — Vals AI harness · Vals AI
  • 88.9%Original — Mercor single-shot · Mercor
Potentially non-equivalent · Evaluation details for MMLU-Pro

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

GPT-5.6 Sol

GPT-6 Sol

Record for GPT-6 Sol on MMLU-Pro Original — Mercor single-shot
  • 83.6%
  • 83.0%
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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on MMMU-Pro 1,730 questions — Mercor single-shot
  • 29.8%low
  • 49.7%medium
  • 56.5%high
  • 60.9%xhigh
  • 66.2%max
  • 43.9%low
  • 54.0%medium
  • 58.3%high
  • 60.5%xhigh
  • 64.4%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 Sol

Record for GPT-5.6 Sol on OSWorld 2.0 v2026.08.08 Offline — Partial

GPT-6 Sol

Record for GPT-6 Sol on OSWorld 2.0 v2026.08.08 Offline — Partial
Record for GPT-6 Sol on OSWorld 2.0 v2026.08.08 Offline — Partial
Record for GPT-6 Sol on OSWorld 2.0 v2026.08.08 Offline — Partial
  • 74.4%
  • 64.8%
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.

GPT-5.6 Sol

Record for GPT-5.6 Sol on Riemann-bench October 2026 leaderboard

GPT-6 Sol

Record for GPT-6 Sol on Riemann-bench October 2026 leaderboard
  • 69.7%
  • 60.7%
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.

SWE AtlasPotentially non-equivalent: Benchmark versions differ. Evaluator sets differ.
  • 46.0%Codebase QnA — Public; Codex · Scale AI
  • 44.1%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.

  • 85.8%
  • 83.1%
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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on Terminal-Bench 2.1 — Vals AI Terminus 2
  • 37.3%
  • 49.4%
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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on Terminal-Bench 4.0 — Codex
  • 37.9%
  • 44.4%
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 Sol

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

GPT-6 Sol

Record for GPT-6 Sol on Terminal-Bench 4.0 — Vals AI mini-swe-agent
Terminal-BenchPotentially non-equivalent: Benchmark versions differ. Evaluator sets differ.
  • 88.8%2.1 · OpenAI
  • 43.8%4.0 — Mercor mini-swe-agent · Mercor
Potentially non-equivalent · Evaluation details for Terminal-Bench

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

GPT-5.6 Sol

Record for GPT-5.6 Sol on Terminal-Bench 2.1

GPT-6 Sol

Record for GPT-6 Sol on Terminal-Bench 4.0 — Mercor mini-swe-agent
  • 58.0%
  • 57.5%
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 Sol

Record for GPT-5.6 Sol on Vals Index 2.1

GPT-6 Sol

Record for GPT-6 Sol on Vals Index 2.1
  • 80.5%
  • 87.8%
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.