Shared benchmarks

19
Shared benchmarks for GPT-5.2 and GPT-5.4
BenchmarkGPT-5.2GPT-5.4
  • 77.9%pro xhigh
  • 65.8%xhigh
  • 82.7%
Evaluation details for BrowseComp

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

GPT-5.2

Record for GPT-5.2 on BrowseComp BrowseComp
Record for GPT-5.2 on BrowseComp BrowseComp

GPT-5.4

Record for GPT-5.4 on BrowseComp BrowseComp
  • 74.1%pro xhigh
  • 70.9%heavy
  • 83.0%
Evaluation details for GDPval

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

GPT-5.2

Record for GPT-5.2 on GDPval GDPval
Record for GPT-5.2 on GDPval GDPval

GPT-5.4

Record for GPT-5.4 on GDPval GDPval
  • 93.2%pro xhigh
  • 92.4%xhigh
  • 92.8%
Evaluation details for GPQA Diamond

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

GPT-5.2

Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond
Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond

GPT-5.4

Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond
  • 73.2%none
  • 82.7%low
  • 87.9%medium
  • 88.2%high
  • 91.4%xhigh
  • 74.7%none
  • 84.8%low
  • 88.9%medium
  • 89.9%high
  • 93.3%xhigh
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-5.2

Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness

GPT-5.4

Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond — Epoch AI harness
  • 91.7%
  • 91.7%
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.

GPT-5.2

Record for GPT-5.2 on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness

GPT-5.4

Record for GPT-5.4 on Graduate-Level Google-Proof Q&A Diamond — Vals AI harness
  • 27.8%
  • 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.

GPT-5.2

GPT-5.4

Record for GPT-5.4 on Humanity's Last Exam Full set — Scale AI evaluation
  • 28.5%
  • 36.5%
Evaluation details for Humanity's Last Exam Text only — Scale AI evaluation

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

GPT-5.2

GPT-5.4

  • 67.6%
  • 70.6%
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.

GPT-5.2

Record for GPT-5.2 on MCP Atlas Public, April 2026 update

GPT-5.4

Record for GPT-5.4 on MCP Atlas Public, April 2026 update
  • 86.2%
  • 87.5%
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.

GPT-5.2

Record for GPT-5.2 on MMLU-Pro Original — Vals AI harness

GPT-5.4

Record for GPT-5.4 on MMLU-Pro Original — Vals AI harness
  • 79.5%
  • 81.2%
Evaluation details for MMMU-Pro No tools

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

GPT-5.2

Record for GPT-5.2 on MMMU-Pro No tools

GPT-5.4

Record for GPT-5.4 on MMMU-Pro No tools
  • 86.7%
  • 87.5%
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.

GPT-5.2

Record for GPT-5.2 on MMMU-Pro Vals AI harness

GPT-5.4

Record for GPT-5.4 on MMMU-Pro Vals AI harness
  • 37.6%
  • 41.6%
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-5.2

Record for GPT-5.2 on Riemann-bench October 2026 leaderboard

GPT-5.4

Record for GPT-5.4 on Riemann-bench October 2026 leaderboard
  • 32.8%low
  • 32.7%medium
  • 34.3%high
  • 37.1%xhigh
  • 45.1%
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.

  • 73.8%
  • 76.9%
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.

GPT-5.2

Record for GPT-5.2 on SWE-bench Verified — Epoch AI harness

GPT-5.4

Record for GPT-5.4 on SWE-bench Verified — Epoch AI harness
  • 75.8%
  • 78.2%
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.

GPT-5.2

Record for GPT-5.2 on SWE-bench Verified — Vals AI mini-swe-agent

GPT-5.4

Record for GPT-5.4 on SWE-bench Verified — Vals AI mini-swe-agent
  • 55.6%
  • 57.7%
Evaluation details for SWE-bench Pro Public

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

GPT-5.2

Record for GPT-5.2 on SWE-bench Pro Public

GPT-5.4

Record for GPT-5.4 on SWE-bench Pro Public
  • 51.7%
  • 58.4%
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.

GPT-5.2

Record for GPT-5.2 on Terminal-Bench 2.0 — Vals AI Terminus 2

GPT-5.4

Record for GPT-5.4 on Terminal-Bench 2.0 — Vals AI Terminus 2
  • 53.5%
  • 67.4%
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.

  • 32.2%
  • 39.4%
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.