Model record

GPT-5.4 mini

xhigh

OpenAI · released Mar 17, 2026

Overall standing

Overall
49.7rank 29 of 62
49.7 is at approximately the 54th percentile of the ranked Overall field. The field spans 19.0 to 62.2, with a median of 48.9.
8 of 8 measured
Category standingsReasoning 64.2Coding 54.5Math 10.0Agentic 21.4
Reasoning
64.2rank 33 of 62
64.2 is at approximately the 48th percentile of the ranked Reasoning field. The field spans 31.6 to 73.3, with a median of 64.8.
4 of 4 measured
Coding
54.5rank 25 of 46
54.5 is at approximately the 47th percentile of the ranked Coding field. The field spans 10.4 to 72.4, with a median of 56.1.
2 of 2 measured
Math
10.0rank 24 of 62
10.0 is at approximately the 61th percentile of the ranked Math field. The field spans 0.0 to 32.3, with a median of 5.3.
1 of 1 measured
Agentic
21.4rank 20 of 46
21.4 is at approximately the 58th percentile of the ranked Agentic field. The field spans 3.3 to 33.4, with a median of 19.6.
1 of 1 measured
Model factsClosed weights · $0.75 / $4.5 per 1M

Copied from the linked provider page; no separate retrieval date is stored.

Provider
OpenAI
Released
Weights
Closed weights
Context window
400,000 tokens
Input price / Mtok
$0.75
Output price / Mtok
$4.5
Reasoning effort
xhigh effort

Scores and sources

Measured scores retain their source and retrieval date. Missing scores are not treated as zero.

Benchmark scores, evaluation settings, sources, and retrieval dates for GPT-5.4 mini
BenchmarkCategoryScoreEvaluation settingsSource
Terminal-Bench v2.1Coding59.2Artificial Analysis independent run; 89 tasks; Terminus 2 on E2B; 3 repeats; pass@1; 250-episode cap; 2-hour timeoutOpen source (opens in a new tab)
τ³-BankingAgentic21.4Artificial Analysis independent run; 97 tasks; 5 repeats; backend-state pass@1; BM25 plus grep retrieval; 200-step capOpen source (opens in a new tab)
AA-LCRReasoning69.3Artificial Analysis independent run; 100 roughly 100k-token multi-document questions; 3 repeats; equality-checker pass@1; no toolsOpen source (opens in a new tab)
Humanity's Last ExamReasoning26.7Artificial Analysis independent run; 2,158 text-only questions; pass@1; GPT-4o (August 2024) equality checker; no toolsOpen source (opens in a new tab)
GPQA DiamondReasoning87.5Artificial Analysis independent run; 198 questions; 5 repeats; regex-graded pass@1; no toolsOpen source (opens in a new tab)
SciCodeCoding49.9Artificial Analysis independent run; 288 test subproblems; scientist background included; 3 repeats; subproblem pass@1Open source (opens in a new tab)
IFBenchReasoning73.3Artificial Analysis independent run; 294 single-turn prompts; 5 repeats; official loose evaluator; prompt-level pass@1Open source (opens in a new tab)
CritPtMath10.0Artificial Analysis independent run; 70 test challenges; 5 repeats; two-step answer parsing; official CritPt grader; pass@1; no toolsOpen source (opens in a new tab)