AIEvals

Models

GPT-5.4

Specifications and API prices for GPT-5.4, recorded by Artificial Analysis.

OpenAIReleased 5 March 2026Context 1,050,000 tokensOutput Not published in the checked sourcePrivate weights$2.50 in / $15.00 out per Mtok

Family GPT-5.416 benchmarks with a published resultNo first place on the independent columns among the 46 models this site listsRead 17 October 2026

Where it stands

Rank compares tracked models with a score. Published shows the source’s rank and field size, when available.

BenchmarkScoreRankPublishedLeaderBehind
Artificial Analysis Intelligence IndexArtificial Analysis3929of 42no list readClaude Opus 5.519 pts
Vibe Code Bench v1.1Vals AI48.47%33of 3852of 106Claude Sonnet 5.543.92 pts
APEX-Agents-AAArtificial Analysis33.26%6of 7no list readGemini 3 5 Flash13.79 pts
τ³-BankingArtificial Analysis39.59%17of 30no list readQwen3.8 Max11.75 pts
IFBenchArtificial Analysis73.95%4of 10no list readGrok 4 20 03098.98 pts
Vals RSI IndexVals AI15.48%20of 2021of 23Claude Opus 5.521.83 pts
τ²-Bench TelecomArtificial Analysis87.13%8of 10no list readJt 35b Flash11.99 pts
CritPtArtificial Analysis23.43%15of 42no list readGPT-5.6 Sol8.86 pts
AA-OmniscienceArtificial Analysis5.7830of 42no list readClaude Opus 5.540.64
AA-LCR v1.1Artificial Analysis82.00%19of 42no list readKimi K36.67 pts
DeepSWE v1.1Datacurve51.77%20of 2022of 28GPT-6 Astra22.35 pts
GDPval-AA v2.1Artificial Analysis1,232.6739of 42no list readClaude Opus 5.5613.5
Humanity's Last ExamArtificial Analysis43.74%23of 42no list readClaude Opus 5.517.61 pts

Nearby on Artificial Analysis Intelligence Index

The two models above and below on the Artificial Analysis Intelligence Index (Artificial Analysis), on every column where this model has a published number.

ModelArtificial Analysis Intelligence IndexindexVibe Code Bench v1.1%
DeepSeek V4.1 FlashDeepSeek3984.74%
Gemini 3.7 FlashGoogle3970.39%
GPT-5.4OpenAI3948.47%
Grok 4.5xAI3969.00%
Claude Sonnet 5Anthropic3881.33%
Bold marks a figure better than this model’s on the same column, whichever model it belongs to. Ranks cover the models shown from the stated source.

Cost and speed, as published

Additional measurements from Artificial Analysis’ own row. Read 30 September 2026.

As publishedCost per taskOutput speed
GPT-5.4 (xhigh)—103.3 tok/s

Results by benchmark

Independent and vendor results retain their own harnesses and effort settings.

BenchmarkIndependent runAs reported by a publisher
APEX-Agents-AAKnowledge work33.26%Artificial Analysisnot published
τ³-BankingTool use39.59%Artificial Analysisnot published
IFBenchInstruction following73.95%Artificial Analysisnot published
Vals RSI IndexAutonomous research15.48%Vals AInot published
τ²-Bench TelecomTool use87.13%Artificial Analysisnot published
CritPtScientific reasoning23.43%Artificial Analysisnot published
AA-OmniscienceFactual knowledge5.78Artificial Analysisnot published
AA-LCR v1.1Long-context reasoning82.00%Artificial Analysisnot published
Terminal-Bench 2.1Agentic coding
78.28%Artificial Analysis
not published
GPQA DiamondScience reasoning
92.02%Artificial Analysis
not published
MMMU ProMultimodal reasoning
78.44%Artificial Analysis
not published
Vibe Code Bench v1.1Product coding48.47%Vals AInot published
DeepSWE v1.1Software engineering51.77%Datacurvenot published
GDPval-AA v2.1Knowledge work1,232.67Artificial Analysisnot published
Humanity's Last ExamMultidisciplinary reasoning43.74%Artificial Analysisnot published
Artificial Analysis Intelligence IndexComposite39Artificial Analysisnot published
Independent column: the suite named under each figure, one harness per suite. Publisher column: each lab’s own table, reproduced on the benchmark’s page. Publisher ranks compare reported scores in the same table row only; ties share a rank and unreported cells are excluded.

Sources

Read on 17 October 2026.