GPT-5.4

OpenAI · openai/gpt-5-4

ProviderOpenAI
Familygpt
Typellm-reasoning
Statusactive
Released2026-03-05
Updated2026-03-05

Reported benchmark scores

These scores carry one collection date for the whole card (2026-04) and a source list rather than a source per score, so they cannot be attributed individually. They are shown as reported and are not verified evidence. Benchmark pages carry the reviewed, dated evidence where it exists.
BenchmarkCatalogue standingScore
Aider Polyglot Benchmarkunassessed72.1
Arena Elo — Codingunassessed1400.0
Arena Elo — Hard Promptsunassessed1507.2
Arena Elo — Mathunassessed1380.0
Arena Elo — Overall (Text)unassessed1390.0
Arena Elo — Style Controlunassessed1483.6
GPQA Diamondunassessed92.8
GraphWalks BFS (256K-1M context)unassessed21.4
Humanity's Last Examunassessed39.8
Humanity's Last Exam (with tools)unassessed52.1
HumanEvalunassessed92.1
IFEvalunassessed91.8
LiveCodeBenchunassessed84.6
MATH-500unassessed95.5
MedQAunassessed93.0
MMLU-Prounassessed84.8
MultiPL-E: C#unassessed88.5
MultiPL-E: Juliaunassessed70.5
MultiPL-E: Kotlin (unconfirmed)unassessed82.8
MultiPL-E: Luaunassessed65.5
MultiPL-E: Perlunassessed62.8
MultiPL-E: PHPunassessed85.2
MultiPL-E: Runassessed68.2
MultiPL-E: Rubyunassessed78.2
MultiPL-E: Scalaunassessed72.5
MultiPL-E: Swiftunassessed76.5
OSWorldunassessed75.0
SWE-bench Agentunassessed55.2
SWE-bench Prounassessed57.7
SWE-bench Verifiedunassessed69.3
τ-benchunassessed58.5
Terminal-Benchunassessed48.8
Terminal-Bench 2.0unassessed75.1
USAMO 2026unassessed95.2

Data

This card as JSON · Edit on GitHub