supermarket

Vision-language

GPT-5.2

By OpenAI, in the GPT-5 family.

Released
11 Dec 2025
Status
Legacy
Context
400K
Max output
128K
Input per 1M tokens
$1.75
Output per 1M tokens
$14.00
Open weights
No
License
proprietary
Parameters
—
Knowledge cutoff
31 Aug 2025
Reasoning
Yes
Modalities
text, image → text

Category scores

Out of 100, from independent and third-party results only, by method category-2026-09-29.1.

Category scores
CategoryScoreRankCoverage
Knowledge & reasoning39.0374 of 4
Coding & software engineering27.4Not ranked1 of 5
Agentic tool use41.1Not ranked1 of 2
Vision & multimodal understanding44.082 of 3
Factuality & hallucination37.1401 of 1
Human preference47.5334 of 5

Benchmark results

Every result we hold for this model, with who ran it and where it was published.

Knowledge & reasoning

Knowledge & reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Chess PuzzlesPuzzles solved49.0%Independentxhigh15 Dec 2025Epoch AI
Chess PuzzlesPuzzles solved40.0%Independenthigh11 Dec 2025Epoch AI
Chess PuzzlesPuzzles solved40.0%Independentmedium11 Dec 2025Epoch AI
Chess PuzzlesPuzzles solved23.0%Independentlow11 Dec 2025Epoch AI
Chess PuzzlesPuzzles solved4.0%Independentnone13 Jul 2026Epoch AI
DTBench (Decision Theory Benchmark)Accuracy (capability questions)90.9%Third partyxhigh—https://conceptualreasoning.ai/dtbench via Epoch AI
Earthborne Rangers (EBR-bench)Objectives reached23.0%Independentxhigh26 Jun 2026Epoch AI
LMCA (Language Model Conceptual Argumentation)Correlation with expert ratings (Pearson r)0.4391Third partyxhigh—https://conceptualreasoning.ai/lmca via Epoch AI
Mystery Game PuzzlesPuzzles solved23.0%Independenthigh6 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved22.0%Independentmedium28 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved14.0%Independentnone27 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved10.0%Independentlow27 Aug 2026Epoch AI
SimpleBenchScore (AVG@5)45.8%Third partyunknown—SimpleBench Leaderboard via Epoch AI
SimpleBenchScore (AVG@5)45.8%Third partyhigh—SimpleBench Leaderboard via Epoch AI

Scientific reasoning

Scientific reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GPQA DiamondAccuracy91.4%Independentxhigh13 Dec 2025Epoch AI
GPQA DiamondAccuracy88.2%Independenthigh11 Dec 2025Epoch AI
GPQA DiamondAccuracy87.9%Independentmedium11 Dec 2025Epoch AI
GPQA DiamondAccuracy82.7%Independentlow11 Dec 2025Epoch AI
GPQA DiamondAccuracy73.2%Independentnone13 Jul 2026Epoch AI

Mathematics

Mathematics results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
FrontierMath Tier 4Accuracy31.7%Independentxhigh11 Jun 2026Epoch AI
FrontierMath Tier 4 (v1)Accuracy18.8%Independenthigh11 Dec 2025Epoch AI
FrontierMath Tier 4 (v1)Accuracy18.8%Independentxhigh14 Dec 2025Epoch AI
FrontierMath Tier 4 (v1)Accuracy16.7%Independentmedium11 Dec 2025Epoch AI
FrontierMath Tier 4 (v1)Accuracy6.3%Independentlow11 Dec 2025Epoch AI
FrontierMath Tiers 1-3Accuracy67.4%Independentxhigh11 Jun 2026Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy40.7%Independentxhigh13 Dec 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy40.3%Independenthigh11 Dec 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy36.9%Independentmedium11 Dec 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy26.6%Independentlow11 Dec 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy96.1%Independenthigh11 Dec 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy96.1%Independentxhigh13 Dec 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy93.9%Independentmedium11 Dec 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy78.9%Independentlow11 Dec 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy62.2%Independentnone13 Jul 2026Epoch AI

Coding & software engineering

Coding & software engineering results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GSOOPT@127.4%Third partyhigh—https://gso-bench.github.io/index.html via Epoch AI
METR Time HorizonsAverage task success (Epoch's column)75.3%Source not statedhigh—Epoch AI via Epoch AI
SWE-bench Verified% resolved73.8%Independenthigh12 Feb 2026Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)64.9%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)64.9%Third partymedium—Terminal-Bench v2 Leaderboard via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)62.9%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)62.9%Third partymedium—Terminal-Bench v2 Leaderboard via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)60.7%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)54.0%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
WeirdML (v2)Average accuracy72.2%Third partyxhigh—https://htihle.github.io/weirdml.html via Epoch AI
WeirdML (v2)Average accuracy63.4%Third partymedium—https://htihle.github.io/weirdml.html via Epoch AI
WeirdML (v2)Average accuracy49.6%Third partynone—WeirdML Leaderboard via Epoch AI
WeirdML (v2)Average accuracy49.6%Third partylow—WeirdML Leaderboard via Epoch AI

Agentic tool use

Agentic tool use results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Deep Research Bench (FutureSearch)Average score41.1%Third partylow—https://evals.futuresearch.ai/#drb via Epoch AI

Professional knowledge work

Professional knowledge work results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GDPvalWins vs experts49.7%Source not statednone—Epoch AI via Epoch AI

Long context

Long context results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
CL-benchTasks solved18.2%Source not statedunknown—Epoch AI via Epoch AI
CL-benchTasks solved18.1%Source not statedhigh—Epoch AI via Epoch AI

Vision & multimodal understanding

Vision & multimodal understanding results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Furniture AssemblyScore38.3%Independentxhigh10 Sep 2026Epoch AI
Visual Physics Comprehension Test (VPCT)Correct84.0%Third partyxhigh—VPCT leaderboard via Epoch AI
Visual Physics Comprehension Test (VPCT)Correct67.0%Third partyhigh—VPCT leaderboard via Epoch AI

Factuality & hallucination

Factuality & hallucination results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
SimpleQA VerifiedAccuracy (% correct)37.1%Independentxhigh27 Aug 2026Epoch AI
SimpleQA VerifiedAccuracy (% correct)34.3%Independenthigh27 Aug 2026Epoch AI
SimpleQA VerifiedAccuracy (% correct)32.8%Independentlow27 Aug 2026Epoch AI
SimpleQA VerifiedAccuracy (% correct)32.7%Independentmedium27 Aug 2026Epoch AI

Human preference

Human preference results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Code Arena: WebDev (arena.ai)Arena score (Elo scale)1415Independent—25 Sep 2026Arena
Code Arena: WebDev (arena.ai)Votes1083Independent—25 Sep 2026Arena
Document Arena (arena.ai)Arena score (Elo scale)1425Independent—13 Sep 2026Arena
Document Arena (arena.ai)Arena score (Elo scale)1423Independenthigh13 Sep 2026Arena
Document Arena (arena.ai)Votes28212Independent—13 Sep 2026Arena
Document Arena (arena.ai)Votes7108Independenthigh13 Sep 2026Arena
Text Arena (arena.ai)Arena score (Elo scale)1437Independenthigh25 Sep 2026Arena
Text Arena (arena.ai)Arena score (Elo scale)1436Independent—25 Sep 2026Arena
Text Arena (arena.ai)Votes83555Independent—25 Sep 2026Arena
Text Arena (arena.ai)Votes49570Independenthigh25 Sep 2026Arena
Vision Arena (arena.ai)Arena score (Elo scale)1243Independenthigh28 Sep 2026Arena
Vision Arena (arena.ai)Arena score (Elo scale)1229Independent—28 Sep 2026Arena
Vision Arena (arena.ai)Votes20149Independent—28 Sep 2026Arena
Vision Arena (arena.ai)Votes19051Independenthigh28 Sep 2026Arena

Other GPT-5 versions

Sources