priorsBuilder

xAI · released 2026-07-16

Grok 4.5

Measured on 91 benchmarks across 3 reasoning efforts.

General frontier

40.7

#27 of 32 models · best at medium

Reasoning efforts in your mix

medium40.7#82 of 128
high40.2#85 of 128
low36.2#103 of 128

Your mix’s benchmarks

8 of 8 measured

reasoning

Humanity's Last Exam

587 results
  1. #61high of 58742.7%

Leader: Opus 5.5 · 61.4%

knowledge

AA-Omniscience

526 results
  1. #51high of 52625.3

Leader: Opus 5.5 · 46.4

reasoning

ARC-AGI-3

57 results
  1. #42medium of 570.3%
  2. #44high of 570.3%
  3. #45low of 570.3%

Leader: GPT-6 Astra · 99.9%

reasoning

BullshitBench v2

196 results
  1. #65high of 19655.0%
  2. #68low of 19654.0%

Leader: Opus 4.8 · 95.0%

coding

DeepSWE v1.1

70 results
  1. #45high of 7053.8%

Leader: GPT-6 Astra · 74.1%

math

FrontierMath Tier 4 (v2)

70 results
  1. #45high of 7024.4%

Leader: GPT-6.1 Sol · 100.0%

reasoning

SimpleBench

97 results
  1. #20high of 9770.0%

Leader: Opus 5.5 · 88.4%

agentic

Terminal-Bench 4.0

35 results
  1. #33high of 3512.4%

Leader: Opus 5.5 · 64.8%

All 91 benchmarks

show 83 more

agentic

AA-AnalystAgent

44 results
  1. #24high of 4435.0%

Leader: Gemini 3.7 Flash · 60.0%

agentic

AA-Briefcase v1.1

224 results
  1. #71high of 2241281

Leader: Sonnet 5.5 · 1823

agentic

Agent Arena

50 results
  1. #28high of 501.2%

Leader: Opus 5.5 · 15.5%

agentic

AutomationBench-AA

219 results
  1. #43high of 21957.9%

Leader: Gemini 4 · 77.5%

agentic

EnterpriseOps-Gym-AA

50 results
  1. #25high of 5040.8%

Leader: Fable 5 · 51.1%

agentic

Harvey's Legal Agent Benchmark

79 results
  1. #11high of 7912.9%

Leader: Muse Spark 1.2 · 25.4%

agentic

Legal Research Bench

78 results
  1. #38high of 7838.0%

Leader: Muse Spark 1.3 · 55.3%

agentic

PostTrainBench v1.1

12 results
  1. #10high of 1223.4%

Leader: Fable 5 · 41.8%

agentic

SkillsBench

34 results
  1. #2high of 3466.0%

Leader: DeepSeek V4.1 Flash · 69.8%

agentic

Tax Agent Bench

69 results
  1. #36high of 6924.9%

Leader: Fable 5.1 · 49.2%

agentic

Terminal-Bench 2.1

18 results
  1. #9high of 1879.3%

Leader: GPT-6 Astra · 87.4%

agentic

Terminal-Bench 4.0 (Vals)

47 results
  1. #37high of 478.6%

Leader: Opus 5.5 · 65.2%

agentic

Time Horizon Index: KSP

13 results
  1. #11high of 136.3%

Leader: Opus 5.5 · 91.3%

agentic

Vending-Bench 2

67 results
  1. #41high of 67$3,887

Leader: GPT-6 Astra · $15,515

agentic

Web Search Index

4 results
  1. #4high of 438.7%

Leader: Fable 5 · 48.5%

agentic

τ³-Banking

206 results
  1. #24high of 20642.1%

Leader: Qwen 3.8 Max · 51.3%

coding

Arena Image-to-WebDev

58 results
  1. #25high of 581564

Leader: Opus 5.5 · 1749

coding

Arena WebDev

122 results
  1. #37high of 1221553

Leader: Opus 5.5 · 1813

coding

Code Migration

77 results
  1. #34high of 7736.6%

Leader: Sonnet 5.5 · 69.8%

coding

FrontierCode 1.1 Extended

130 results
  1. #52high of 13056.5%
  2. #57medium of 13056.1%
  3. #74low of 13053.1%

Leader: Opus 5.5 · 65.3%

coding

FrontierCode 1.1 Main

130 results
  1. #55high of 13042.4%
  2. #61medium of 13041.9%
  3. #78low of 13037.9%

Leader: Opus 5.5 · 54.6%

coding

FrontierSWE v1

17 results
  1. #4high of 1772.1%

Leader: Fable 5 · 88.2%

coding

IOI

44 results
  1. #34high of 4440.6%

Leader: Gemini 4 · 100.0%

coding

LiveCodeBench (Vals)

127 results
  1. #15high of 12787.4%

Leader: Fable 5.1 · 90.5%

coding

ProgramBench v1 Almost Resolved

62 results
  1. #36high of 624.5%

Leader: Opus 5.5 · 65.0%

coding

ProgramBench v1 Fully Resolved

62 results
  1. #48high of 620.0%

Leader: Opus 5.5 · 18.5%

coding

ProgramBench v1 Raw Pass Rate

62 results
  1. #30high of 6265.2%

Leader: Opus 5.5 · 87.0%

coding

SciCode

226 results
  1. #54high of 22655.0%

Leader: Opus 5.5 · 66.9%

coding

SWE-bench Verified (Vals)

85 results
  1. #13high of 8586.6%

Leader: Opus 5 · 97.0%

coding

Terminal-Bench 2.1 (Vals)

74 results
  1. #31high of 7467.8%

Leader: Opus 5.5 · 87.6%

coding

Terminal-Bench v2.1-AA

238 results
  1. #40high of 23881.6%

Leader: Fable 5.1 · 91.4%

coding

Terminal-Bench v4-AA

216 results
  1. #85high of 21610.6%

Leader: Sonnet 5.5 · 63.6%

coding

Vibe Code Bench v1.1

105 results
  1. #40high of 10569.0%

Leader: Sonnet 5.5 · 92.4%

coding

WeirdML v2

127 results
  1. #70high of 12746.4%

Leader: GPT-6 Astra Pro · 93.6%

composite

AA Intelligence Index v4.3.2

610 results
  1. #71high of 61038.8

Leader: Opus 5.5 · 57.6

composite

Arena Text

220 results
  1. #46high of 2201466

Leader: Gemini 4 · 1525

composite

Arena Text — Multi-Turn

220 results
  1. #49high of 2201472

Leader: Gemini 4 · 1553

composite

Vals Index

47 results
  1. #39high of 4744.7%

Leader: Gemini 4 · 68.9%

composite

Vals Multimodal Index v1.2

33 results
  1. #15high of 3363.4%

Leader: Fable 5 · 74.2%

cost

AA Intelligence Index Cost per Task

177 results
  1. #107high of 177$0.88

Leader: GPT-6 Luna · $0.00

cost

Cost to Run AA Intelligence Index

177 results
  1. #110high of 177$1,244

Leader: GPT-6 Luna · $10.63

creative

MineBench

72 results
  1. #34high of 721584

Leader: GPT-6 Astra Pro · 2281

creative

VoxelBench (text)

55 results
  1. #23high of 551707

Leader: GPT-6 Astra · 2732

economics

CorpFin (Vals)

117 results
  1. #13high of 11767.4%

Leader: Opus 5 · 73.2%

economics

Excel Modeling Benchmark

74 results
  1. #50high of 7452.9%

Leader: Fable 5.1 · 76.7%

economics

Finance Agent v2

78 results
  1. #48high of 7848.3%

Leader: Gemini 4 · 65.4%

economics

GDP.pdf

210 results
  1. #68high of 21018.8%

Leader: Muse Wonder Full (Max) · 32.6%

economics

GDPval-AA v2.1

288 results
  1. #76high of 2881386

Leader: Opus 5.5 · 1866

economics

MortgageTax (Vals)

84 results
  1. #55high of 8461.8%

Leader: Opus 5 · 72.1%

economics

TaxEval (Vals)

125 results
  1. #72high of 12571.7%

Leader: Muse Spark 1.2 · 80.4%

knowledge

AA-Omniscience Accuracy

526 results
  1. #54high of 52651.5%

Leader: Fable 5.1 · 67.2%

knowledge

AA-Omniscience Non-hallucination

526 results
  1. #136high of 52645.9%

Leader: MiniCPM5-1B (Non-reasoning) · 99.1%

knowledge

Arena Search

28 results
  1. #7high of 281213

Leader: GPT-5.6 Sol · 1257

knowledge

GPQA Diamond

539 results
  1. #26high of 53993.1%

Leader: GPT-6 Astra · 96.3%

knowledge

GPQA Diamond (Vals)

122 results
  1. #13high of 12292.9%

Leader: Gemini 3.1 Pro · 95.5%

knowledge

MedCode

94 results
  1. #43high of 9443.3%

Leader: Opus 5 · 63.6%

knowledge

MedScribe

96 results
  1. #18high of 9686.9%

Leader: Opus 5.5 · 91.4%

knowledge

MMLU Pro (Vals)

122 results
  1. #14high of 12289.2%

Leader: Fable 5.1 · 92.4%

knowledge

MMMU Pro (Vals)

81 results
  1. #72high of 8161.8%

Leader: Fable 5.1 · 90.6%

knowledge

MMMU-Pro

265 results
  1. #47high of 26580.4%

Leader: Opus 5.5 · 87.7%

long-context

AA-LCR v1.1

523 results
  1. #95high of 52379.3%

Leader: Kimi K3 · 88.7%

long-context

Arena Document

43 results
  1. #26high of 431452

Leader: Opus 5 · 1516

long-context

MLCR-AA

94 results
  1. #50high of 9415.0%

Leader: Sonnet 5.5 · 75.0%

math

FrontierMath Tiers 1–3 (v2)

115 results
  1. #40high of 11557.2%

Leader: GPT-6.1 Sol · 93.7%

math

ProofBench v1.1

52 results
  1. #38high of 5231.0%

Leader: Sonnet 5.5 · 100.0%

other

Arena Vision

116 results
  1. #26high of 1161280

Leader: Fable 5 · 1308

reasoning

ARC-AGI-1

215 results
  1. #83medium of 21587.2%
  2. #97high of 21585.7%
  3. #107low of 21579.2%

Leader: Fable 5 · 98.5%

reasoning

ARC-AGI-2

216 results
  1. #94high of 21652.6%
  2. #95medium of 21652.6%
  3. #112low of 21633.1%

Leader: GPT-6 Astra · 95.0%

reasoning

Blueprint-Bench 2

28 results
  1. #20high of 2827.3%

Leader: Gemini 4 · 54.4%

reasoning

CritPt

529 results
  1. #86high of 52915.4%

Leader: Muse Wonder Full (Max) · 32.9%

reasoning

Furniture Assembly

31 results
  1. #28high of 3122.5%

Leader: Opus 5.5 · 83.3%

reasoning

LegalBench

135 results
  1. #18high of 13586.0%

Leader: Fable 5 · 88.6%

reasoning

Roboflow Visual Reasoning

102 results
  1. #63high of 10259.8%
  2. #69low of 10257.6%

Leader: GPT-6 Astra · 91.2%

reasoning

SAGE

80 results
  1. #67high of 8035.0%

Leader: Opus 4.7 · 56.1%

reasoning

SimpleQA Verified

47 results
  1. #12high of 4753.8%

Leader: Gemini 3.1 Pro · 77.5%

reliability

Tool-call error rate

171 results
  1. #23high of 1710.3%

Leader: GPT-5.4 Pro · 0.0%

reliability

Uptime (7d)

181 results
  1. #127high of 18198.6%

Leader: GPT-5 Pro · 100.0%

safety

CyberBench v1.1

45 results
  1. #20high of 4571.0%

Leader: Wonder · 80.4%

safety

SRE Bench

32 results
  1. #27high of 320.8%

Leader: GPT-6 Astra · 56.9%

speed

Latency (Time to First Token)

351 results
  1. #264high of 35114s

Leader: Gemini 2.5 Flash-Lite · 0.3s

speed

Output Speed (tokens/s)

351 results
  1. #261high of 35159.3

Leader: Celeris-1 · 1522.5

speed

Provider latency

177 results
  1. #117high of 1771.8s

Leader: Qwen 2.5 7B · 0.2s

speed

Provider throughput

176 results
  1. #104high of 17653.5

Leader: Muse Spark 1.1 · 204.3