supermarket

Vision-language

GPT-5.4 nano

By OpenAI, in the GPT-5 family.

Released
17 Mar 2026
Status
Legacy
Context
400K
Max output
128K
Input per 1M tokens
$0.20
Output per 1M tokens
$1.25
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 & reasoning34.9432 of 4
Factuality & hallucination11.7661 of 1
Human preference37.1Not 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%Independenthigh14 Apr 2026Epoch AI
Chess PuzzlesPuzzles solved17.0%Independentlow7 Aug 2026Epoch AI
Chess PuzzlesPuzzles solved3.0%Independentnone7 Aug 2026Epoch AI
DTBench (Decision Theory Benchmark)Accuracy (capability questions)80.3%Third partyxhigh—https://conceptualreasoning.ai/dtbench via Epoch AI
LMCA (Language Model Conceptual Argumentation)Correlation with expert ratings (Pearson r)0.3686Third partyxhigh—https://conceptualreasoning.ai/lmca via Epoch AI
Mystery Game PuzzlesPuzzles solved9.0%Independentnone27 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved6.0%Independentmedium27 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved5.0%Independenthigh27 Aug 2026Epoch AI
Mystery Game PuzzlesPuzzles solved3.0%Independentlow27 Aug 2026Epoch AI

Scientific reasoning

Scientific reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GPQA DiamondAccuracy78.5%Independenthigh14 Apr 2026Epoch AI
GPQA DiamondAccuracy72.2%Independentlow7 Aug 2026Epoch AI
GPQA DiamondAccuracy55.6%Independentnone7 Aug 2026Epoch AI

Mathematics

Mathematics results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
FrontierMath Tier 4Accuracy12.2%Independenthigh12 Jun 2026Epoch AI
FrontierMath Tier 4 (v1)Accuracy6.3%Independenthigh15 Apr 2026Epoch AI
FrontierMath Tiers 1-3Accuracy44.9%Independenthigh12 Jun 2026Epoch AI
FrontierMath Tiers 1-3Accuracy20.4%Independentlow28 Aug 2026Epoch AI
FrontierMath Tiers 1-3Accuracy4.6%Independentnone28 Aug 2026Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy25.9%Independenthigh15 Apr 2026Epoch AI
OTIS Mock AIME 2024-2025Accuracy87.8%Independenthigh14 Apr 2026Epoch AI
OTIS Mock AIME 2024-2025Accuracy68.9%Independentlow7 Aug 2026Epoch AI
OTIS Mock AIME 2024-2025Accuracy46.7%Independentnone7 Aug 2026Epoch AI

Coding & software engineering

Coding & software engineering results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
WeirdML (v2)Average accuracy49.2%Third partyhigh—https://htihle.github.io/weirdml.html via Epoch AI
WeirdML (v2)Average accuracy38.0%Third partynone—https://htihle.github.io/weirdml.html via Epoch AI

Factuality & hallucination

Factuality & hallucination results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
SimpleQA VerifiedAccuracy (% correct)11.7%Independenthigh27 Aug 2026Epoch AI

Human preference

Human preference results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Text Arena (arena.ai)Arena score (Elo scale)1401Independenthigh25 Sep 2026Arena
Text Arena (arena.ai)Votes62171Independenthigh25 Sep 2026Arena
Vision Arena (arena.ai)Arena score (Elo scale)1200Independenthigh28 Sep 2026Arena
Vision Arena (arena.ai)Votes25685Independenthigh28 Sep 2026Arena

Other GPT-5 versions

Sources