AIEvals

Models

GPT-5.5

Specifications and API prices for GPT-5.5, recorded by Artificial Analysis.

OpenAIReleased 23 April 2026Context 922,000 tokensOutput Not published in the checked sourcePrivate weights$5.00 in / $30.00 out per Mtok

Family GPT-5.544 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 Analysis3833of 42no list readClaude Opus 5.520 pts
Terminal-Bench 4.0Artificial Analysis14.65%27of 40no list readClaude Sonnet 5.548.99 pts
SWE-bench (Verified)Vals AI82.60%15of 2319of 88Claude Opus 514.40 pts
LiveCodeBenchVals AI85.30%15of 2135of 143Claude Fable 5.15.22 pts
GPQA DiamondVals AI93.18%9of 2310of 138Gemini 3.1 Pro2.27 pts
MMLU ProVals AI88.14%14of 2421of 138Claude Fable 5.14.24 pts
MMMU ProVals AI88.27%7of 208of 93Claude Fable 5.12.37 pts
Vibe Code Bench v1.1Vals AI69.85%30of 3834of 106Claude Sonnet 5.522.54 pts
ProgramBenchVals AI0.50%13of 2913of 54Claude Opus 5.518.00 pts
APEX-Agents-AAArtificial Analysis37.68%4of 7no list readGemini 3 5 Flash9.37 pts
τ³-BankingArtificial Analysis38.97%19of 30no list readQwen3.8 Max12.37 pts
ITBench-AAArtificial Analysis45.81%12of 18no list readGPT-5.6 Sol10.40 pts
IFBenchArtificial Analysis75.85%3of 10no list readGrok 4 20 03097.08 pts
AA-AnalystAgentArtificial Analysis50.00%5of 18no list readGemini 3.7 Flash10.00 pts
MLCR-AAArtificial Analysis17.78%24of 37no list readClaude Sonnet 5.557.22 pts
Vals RSI IndexVals AI16.48%19of 2019of 23Claude Opus 5.520.83 pts
τ²-Bench TelecomArtificial Analysis93.86%5of 10no list readJt 35b Flash5.26 pts
Harvey LAB-AAArtificial Analysis86.28%17of 20no list readMuse Spark 1.3 Max9.20 pts
EnterpriseOps-Gym-AAArtificial Analysis46.64%9of 22no list readClaude Fable 54.48 pts
AutomationBench-AAArtificial Analysis47.30%32of 40no list readGemini 4 Argon30.21 pts
SciCodeArtificial Analysis55.79%20of 40no list readClaude Opus 5.511.11 pts
GDP.pdfArtificial Analysis21.20%19of 40no list readGPT-6.1 Sol9.80 pts
CritPtArtificial Analysis27.14%11of 42no list readGPT-5.6 Sol5.15 pts
AA-OmniscienceArtificial Analysis20.5221of 42no list readClaude Opus 5.525.9
AA-LCR v1.1Artificial Analysis84.33%6of 42no list readKimi K34.34 pts
Terminal-Bench 2.1Vals AI76.40%16of 3516of 75Claude Opus 5.511.24 pts
LegalBenchVals AI86.52%9of 2712of 149Claude Fable 52.04 pts
Harvey's Legal Agent BenchmarkVals AI3.75%27of 3734of 73Muse Spark 1.221.67 pts
SkillsBenchVals AI62.21%4of 205of 35DeepSeek V4.1 Flash7.59 pts
Vals Multimodal IndexVals AI68.07%9of 179of 33Claude Fable 56.08 pts
Code MigrationVals AI45.16%16of 3716of 71Claude Sonnet 5.524.67 pts
EMBVals AI64.54%22of 3723of 69Claude Fable 5.112.13 pts
Finance Agent (v2)Vals AI51.76%27of 3733of 73Gemini 4 Argon13.64 pts
Legal Research BenchVals AI40.38%23of 3725of 72Muse Spark 1.3 Max14.91 pts
MedCodeVals AI49.10%13of 3524of 104Claude Opus 514.47 pts
MedScribeVals AI86.87%16of 3618of 106Claude Opus 5.54.56 pts
Public Benefits Bench v1.1Vals AI60.89%24of 2928of 45Claude Opus 516.04 pts
SAGEVals AI51.53%8of 3213of 90Claude Opus 4.74.57 pts
SRE BenchVals AI3.82%11of 1511of 16GPT-6 Astra53.05 pts
Tax Agent BenchVals AI60.46%27of 3731of 64Claude Fable 5.117.18 pts
DeepSWE v1.1Datacurve67.04%11of 2011of 28GPT-6 Astra7.08 pts
GDPval-AA v2.1Artificial Analysis1,335.8937of 42no list readClaude Opus 5.5510.28
AA-Briefcase v1.1Artificial Analysis1,137.1836of 41no list readClaude Opus 5.5684.67
Humanity's Last ExamArtificial Analysis45.78%19of 42no list readClaude Opus 5.515.57 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 IndexindexTerminal-Bench 4.0%SWE-bench%LiveCodeBench%GPQA Diamond%MMLU Pro%MMMU Pro%Vibe Code Bench v1.1%ProgramBench%
Grok 4.5xAI3910.61%86.60%87.35%92.93%89.22%61.79%69.00%0.00%
Claude Sonnet 5Anthropic3814.14%79.60%82.43%88.89%87.55%83.01%81.33%0.00%
GPT-5.5OpenAI3814.65%82.60%85.30%93.18%88.14%88.27%69.85%0.50%
MiMo-V2.6-FlashXiaomi3822.73%–––––78.96%0.50%
GPT-5.6 LunaOpenAI3711.62%93.00%–91.67%86.04%85.03%77.06%0.00%
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-5.5 (xhigh)$2.6393.1 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 work37.68%Artificial Analysisnot published
τ³-BankingTool use38.97%Artificial Analysisnot published
ITBench-AAIT operations45.81%Artificial Analysisnot published
IFBenchInstruction following75.85%Artificial Analysisnot published
AA-AnalystAgentData analysis50.00%Artificial Analysisnot published
MLCR-AAMedical reasoning17.78%Artificial Analysisnot published
Vals RSI IndexAutonomous research16.48%Vals AInot published
τ²-Bench TelecomTool use93.86%Artificial Analysisnot published
Harvey LAB-AALegal reasoning86.28%Artificial Analysisnot published
EnterpriseOps-Gym-AAEnterprise workflows46.64%Artificial Analysisnot published
AutomationBench-AAWorkflow automation47.30%Artificial Analysisnot published
SciCodeScientific coding55.79%Artificial Analysisnot published
GDP.pdfDocument reasoning21.20%Artificial Analysisnot published
CritPtScientific reasoning27.14%Artificial Analysisnot published
AA-OmniscienceFactual knowledge20.52Artificial Analysisnot published
AA-LCR v1.1Long-context reasoning84.33%Artificial Analysisnot published
Terminal-Bench 4.0Agentic coding14.65%Artificial Analysisnot published
Terminal-Bench 2.1Agentic coding76.40%Vals AI
84.27%Artificial Analysis
not published
SWE-bench (Verified)Software engineering82.60%Vals AInot published
LiveCodeBenchCode generation85.30%Vals AInot published
GPQA DiamondScience reasoning93.18%Vals AI
93.54%Artificial Analysis
not published
MMLU ProKnowledge88.14%Vals AInot published
MMMU ProMultimodal reasoning88.27%Vals AI
79.88%Artificial Analysis
not published
Vibe Code Bench v1.1Product coding69.85%Vals AInot published
ProgramBenchAgentic coding0.50%Vals AInot published
LegalBenchProfessional knowledge86.52%Vals AInot published
Harvey's Legal Agent BenchmarkProfessional knowledge3.75%Vals AInot published
SkillsBenchAgentic tooling62.21%Vals AInot published
Vals Multimodal IndexComposite68.07%Vals AInot published
Code MigrationAgentic coding45.16%Vals AInot published
EMBFinancial modelling64.54%Vals AInot published
Finance Agent (v2)Financial analysis51.76%Vals AInot published
Legal Research BenchProfessional knowledge40.38%Vals AInot published
MedCodeHealthcare administration49.10%Vals AInot published
MedScribeHealthcare administration86.87%Vals AInot published
Public Benefits Bench v1.1Public services60.89%Vals AInot published
SAGEEducation51.53%Vals AInot published
SRE BenchSecurity3.82%Vals AInot published
Tax Agent BenchProfessional knowledge60.46%Vals AInot published
DeepSWE v1.1Software engineering67.04%Datacurvenot published
GDPval-AA v2.1Knowledge work1,335.89Artificial Analysisnot published
AA-Briefcase v1.1Knowledge work1,137.18Artificial Analysisnot published
Humanity's Last ExamMultidisciplinary reasoning45.78%Artificial Analysisnot published
Artificial Analysis Intelligence IndexComposite38Artificial 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.