AIEvals

Models

Grok 4.5

Specifications and API prices for Grok 4.5, recorded by Artificial Analysis.

xAIReleased 8 July 2026Context 500,000 tokensOutput Not published in the checked sourcePrivate weights$2.00 in / $6.00 out per Mtok

Family Grok 4.543 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
Vals IndexVals AI44.72%30of 3333of 41Gemini 4 Argon24.18 pts
Artificial Analysis Intelligence IndexArtificial Analysis3929of 42no list readClaude Opus 5.519 pts
Terminal-Bench 4.0Artificial Analysis10.61%35of 40no list readClaude Sonnet 5.553.03 pts
SWE-bench (Verified)Vals AI86.60%12of 2313of 88Claude Opus 510.40 pts
LiveCodeBenchVals AI87.35%11of 2115of 143Claude Fable 5.13.17 pts
GPQA DiamondVals AI92.93%11of 2312of 138Gemini 3.1 Pro2.52 pts
MMLU ProVals AI89.22%10of 2414of 138Claude Fable 5.13.16 pts
MMMU ProVals AI61.79%20of 2083of 93Claude Fable 5.128.85 pts
Vibe Code Bench v1.1Vals AI69.00%31of 3835of 106Claude Sonnet 5.523.39 pts
ProgramBenchVals AI0.00%19of 2923of 54Claude Opus 5.518.50 pts
τ³-BankingArtificial Analysis42.06%13of 30no list readQwen3.8 Max9.28 pts
AA-AnalystAgentArtificial Analysis35.00%15of 18no list readGemini 3.7 Flash25.00 pts
MLCR-AAArtificial Analysis15.00%31of 37no list readClaude Sonnet 5.560.00 pts
ARC-AGI-3ARC Prize · Standard0.32%9of 11no list readGPT-6 Astra62.39 pts
Harvey LAB-AAArtificial Analysis92.42%8of 20no list readMuse Spark 1.3 Max3.06 pts
EnterpriseOps-Gym-AAArtificial Analysis40.82%15of 22no list readClaude Fable 510.30 pts
AutomationBench-AAArtificial Analysis57.93%20of 40no list readGemini 4 Argon19.58 pts
SciCodeArtificial Analysis54.98%21of 40no list readClaude Opus 5.511.92 pts
GDP.pdfArtificial Analysis18.80%25of 40no list readGPT-6.1 Sol12.20 pts
CritPtArtificial Analysis15.43%33of 42no list readGPT-5.6 Sol16.86 pts
AA-OmniscienceArtificial Analysis25.3218of 42no list readClaude Opus 5.521.1
AA-LCR v1.1Artificial Analysis79.33%33of 42no list readKimi K39.34 pts
Terminal-Bench 2.1Vals AI67.79%27of 3532of 75Claude Opus 5.519.85 pts
ProofBench v1.1Vals AI31.00%28of 3432of 46Claude Sonnet 5.569.00 pts
LegalBenchVals AI85.97%12of 2717of 149Claude Fable 52.59 pts
Harvey's Legal Agent BenchmarkVals AI12.92%5of 377of 73Muse Spark 1.212.50 pts
SkillsBenchVals AI66.03%2of 202of 35DeepSeek V4.1 Flash3.77 pts
Vals Multimodal IndexVals AI63.42%13of 1715of 33Claude Fable 510.73 pts
Code MigrationVals AI36.60%27of 3730of 71Claude Sonnet 5.533.23 pts
CyberBench v1.1Vals AI71.01%14of 3117of 43GPT-6 Sol6.97 pts
EMBVals AI52.91%32of 3745of 69Claude Fable 5.123.76 pts
Finance Agent (v2)Vals AI48.35%33of 3743of 73Gemini 4 Argon17.05 pts
IOIVals AI40.56%27of 3129of 38Gemini 4 Argon59.44 pts
Legal Research BenchVals AI37.98%28of 3731of 72Muse Spark 1.3 Max17.31 pts
MedCodeVals AI43.29%26of 3544of 104Claude Opus 520.28 pts
MedScribeVals AI86.88%15of 3617of 106Claude Opus 5.54.55 pts
SAGEVals AI34.97%31of 3270of 90Claude Opus 4.721.13 pts
SRE BenchVals AI0.76%14of 1514of 16GPT-6 Astra56.11 pts
Tax Agent BenchVals AI64.32%18of 3721of 64Claude Fable 5.113.32 pts
DeepSWE v1.1Datacurve53.76%19of 2019of 28GPT-6 Astra20.36 pts
GDPval-AA v2.1Artificial Analysis1,370.4435of 42no list readClaude Opus 5.5475.73
AA-Briefcase v1.1Artificial Analysis1,282.932of 41no list readClaude Opus 5.5538.95
Humanity's Last ExamArtificial Analysis42.68%28of 42no list readClaude Opus 5.518.67 pts

Nearby on Vals Index

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

ModelVals Index%Artificial Analysis Intelligence IndexindexTerminal-Bench 4.0%SWE-bench%LiveCodeBench%GPQA Diamond%MMLU Pro%MMMU Pro%Vibe Code Bench v1.1%ProgramBench%
Qwen3.8 MaxAlibaba48.27%4018.69%85.60%87.85%93.69%88.60%88.04%64.70%0.00%
DeepSeek V4 ProDeepSeek47.63%3614.14%96.40%87.53%92.42%86.97%–82.30%0.00%
Grok 4.5xAI44.72%3910.61%86.60%87.35%92.93%89.22%61.79%69.00%0.00%
MiniMax M3MiniMax36.53%292.02%75.00%82.15%92.68%84.22%81.16%47.57%–
Gemini 3.1 ProGoogle33.44%304.04%78.80%88.49%95.45%90.99%88.21%32.03%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
Grok 4.5 (high)$1.0457.8 tok/s

Results by benchmark

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

BenchmarkIndependent runAs reported by a publisher
τ³-BankingTool use42.06%Artificial Analysisnot published
AA-AnalystAgentData analysis35.00%Artificial Analysisnot published
MLCR-AAMedical reasoning15.00%Artificial Analysisnot published
ARC-AGI-3Interactive reasoning0.32%ARC Prize · Standardnot published
Harvey LAB-AALegal reasoning92.42%Artificial Analysisnot published
EnterpriseOps-Gym-AAEnterprise workflows40.82%Artificial Analysisnot published
AutomationBench-AAWorkflow automation57.93%Artificial Analysisnot published
SciCodeScientific coding54.98%Artificial Analysisnot published
GDP.pdfDocument reasoning18.80%Artificial Analysisnot published
CritPtScientific reasoning15.43%Artificial Analysisnot published
AA-OmniscienceFactual knowledge25.32Artificial Analysisnot published
AA-LCR v1.1Long-context reasoning79.33%Artificial Analysisnot published
Terminal-Bench 4.0Agentic coding10.61%Artificial Analysis
8.59%Vals AI
not published
Terminal-Bench 2.1Agentic coding67.79%Vals AI
81.65%Artificial Analysis
not published
SWE-bench (Verified)Software engineering86.60%Vals AInot published
LiveCodeBenchCode generation87.35%Vals AInot published
GPQA DiamondScience reasoning92.93%Vals AI
93.13%Artificial Analysis
not published
MMLU ProKnowledge89.22%Vals AInot published
MMMU ProMultimodal reasoning61.79%Vals AI
80.40%Artificial Analysis
not published
Vibe Code Bench v1.1Product coding69.00%Vals AInot published
ProgramBenchAgentic coding0.00%Vals AInot published
ProofBench v1.1Mathematics31.00%Vals AInot published
LegalBenchProfessional knowledge85.97%Vals AInot published
Harvey's Legal Agent BenchmarkProfessional knowledge12.92%Vals AInot published
SkillsBenchAgentic tooling66.03%Vals AInot published
Vals Multimodal IndexComposite63.42%Vals AInot published
Vals IndexComposite44.72%Vals AInot published
Code MigrationAgentic coding36.60%Vals AInot published
CyberBench v1.1Security71.01%Vals AInot published
EMBFinancial modelling52.91%Vals AInot published
Finance Agent (v2)Financial analysis48.35%Vals AInot published
IOICompetitive programming40.56%Vals AInot published
Legal Research BenchProfessional knowledge37.98%Vals AInot published
MedCodeHealthcare administration43.29%Vals AInot published
MedScribeHealthcare administration86.88%Vals AInot published
SAGEEducation34.97%Vals AInot published
SRE BenchSecurity0.76%Vals AInot published
Tax Agent BenchProfessional knowledge64.32%Vals AInot published
DeepSWE v1.1Software engineering53.76%Datacurvenot published
GDPval-AA v2.1Knowledge work1,370.44Artificial Analysisnot published
AA-Briefcase v1.1Knowledge work1,282.9Artificial Analysisnot published
Humanity's Last ExamMultidisciplinary reasoning42.68%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.