supermarket

Vision-language

GPT-5 mini

By OpenAI, in the GPT-5 family.

Released
7 Aug 2025
Status
Deprecated
Retired
11 Dec 2026
Context
400K
Max output
128K
Input per 1M tokens
$0.25
Output per 1M tokens
$2.00
Open weights
No
License
proprietary
Parameters
—
Knowledge cutoff
31 May 2024
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 & reasoning33.3472 of 4
Vision & multimodal understanding10.3Not ranked1 of 3
Factuality & hallucination21.6571 of 1
Human preference35.4Not ranked2 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 solved30.0%Independenthigh7 Aug 2026Epoch AI
Chess PuzzlesPuzzles solved12.0%Independentlow7 Aug 2026Epoch AI
Chess PuzzlesPuzzles solved7.0%Independentminimal7 Aug 2026Epoch AI
DTBench (Decision Theory Benchmark)Accuracy (capability questions)80.5%Third partyhigh—https://conceptualreasoning.ai/dtbench via Epoch AI
LMCA (Language Model Conceptual Argumentation)Correlation with expert ratings (Pearson r)0.342Third partyhigh—https://conceptualreasoning.ai/lmca via Epoch AI
Mystery Game PuzzlesPuzzles solved10.0%Independentminimal27 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved5.0%Independenthigh27 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved4.0%Independentmedium27 Aug 2026Epoch AI

Scientific reasoning

Scientific reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GPQA DiamondAccuracy75.0%Independenthigh30 Oct 2025Epoch AI
GPQA DiamondAccuracy71.7%Independentminimal7 Aug 2026Epoch AI
GPQA DiamondAccuracy71.7%Independentmedium7 Aug 2025Epoch AI

Mathematics

Mathematics results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
FrontierMath Tier 4Accuracy12.2%Independenthigh12 Jun 2026Epoch AI
FrontierMath Tier 4 (v1)Accuracy6.3%Independenthigh30 Oct 2025Epoch AI
FrontierMath Tier 4 (v1)Accuracy4.2%Independentmedium7 Aug 2025Epoch AI
FrontierMath Tiers 1-3Accuracy46.7%Independenthigh12 Jun 2026Epoch AI
FrontierMath Tiers 1-3Accuracy18.2%Independentlow27 Aug 2026Epoch AI
FrontierMath Tiers 1-3Accuracy6.0%Independentminimal27 Aug 2026Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy27.2%Independenthigh13 Nov 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy20.3%Independentmedium13 Nov 2025Epoch AI
MATH Level 5Accuracy97.8%Independenthigh30 Oct 2025Epoch AI
MATH Level 5Accuracy96.8%Independentmedium20 Aug 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy86.7%Independenthigh30 Oct 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy78.3%Independentmedium7 Aug 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy55.6%Independentminimal7 Aug 2026Epoch AI

Coding & software engineering

Coding & software engineering results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
SWE-bench Verified% resolved64.7%Independentmedium1 Feb 2026Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)34.8%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)31.9%Third partymedium—Terminal-Bench v2 Leaderboard via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)31.9%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)29.2%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)29.2%Third partymedium—Terminal-Bench v2 Leaderboard via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)24.0%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)24.0%Third partymedium—Terminal-Bench v2 Leaderboard via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)22.2%Third partyunknown—https://www.tbench.ai/leaderboard/terminal-bench/2.0 via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)22.2%Third partymedium—Terminal-Bench v2 Leaderboard via Epoch AI
Terminal-Bench 2.0Accuracy (mean over trials)34.8%Source not statedunknown—Epoch AI via Epoch AI
WeirdML (v2)Average accuracy52.7%Third partyhigh—https://htihle.github.io/weirdml.html via Epoch AI

Long context

Long context results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Fiction.LiveBench% correct at 16K tokens (Epoch's column)69.4%Third partymedium—Fiction.live leaderboard via Epoch AI

Vision & multimodal understanding

Vision & multimodal understanding results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Visual Physics Comprehension Test (VPCT)Correct40.2%Third partymedium—VPCT leaderboard via Epoch AI
Visual Physics Comprehension Test (VPCT)Correct39.0%Third partyhigh—VPCT leaderboard via Epoch AI

Factuality & hallucination

Factuality & hallucination results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
SimpleQA VerifiedAccuracy (% correct)21.6%Independenthigh10 Aug 2026Epoch AI

Writing

Writing results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
LLM Creative Story-Writing Benchmark V1 (Lech Mazur)Mean rubric score (percent of the 10-point scale)83.1%Third partymedium—lechmazur/writing Github repository via Epoch AI

Human preference

Human preference results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Text Arena (arena.ai)Arena score (Elo scale)1390Independenthigh25 Sep 2026Arena
Text Arena (arena.ai)Votes26641Independenthigh25 Sep 2026Arena
Vision Arena (arena.ai)Arena score (Elo scale)1183Independenthigh28 Sep 2026Arena
Vision Arena (arena.ai)Votes28971Independenthigh28 Sep 2026Arena

Other GPT-5 versions

Sources