priorsBuilder

OpenAI · released 2026-09-22

GPT-6 Luna

Measured on 75 benchmarks across 6 reasoning efforts.

General frontier

42.1

#25 of 32 models · best at max

Reasoning efforts in your mix

max42.1#79 of 128
xhigh38.2#93 of 128
high36.3#102 of 128
medium35.4#108 of 128
low28.5#119 of 128
none23.4#123 of 128

Your mix’s benchmarks

6 of 8 measured

reasoning

Humanity's Last Exam

587 results
  1. #95max of 58738.5%
  2. #123xhigh of 58734.3%
  3. #136high of 58732.9%
  4. #161medium of 58728.3%
  5. #206low of 58720.3%
  6. #337none of 5878.6%

Leader: Opus 5.5 · 61.4%

knowledge

AA-Omniscience

526 results
  1. #128max of 5260.7
  2. #142xhigh of 526-1.8
  3. #158medium of 526-5.0
  4. #162high of 526-5.5
  5. #180low of 526-9.2
  6. #242none of 526-21.9

Leader: Opus 5.5 · 46.4

reasoning

ARC-AGI-3

57 results
  1. #34max of 570.6%
  2. #36xhigh of 570.5%
  3. #40high of 570.4%
  4. #43medium of 570.3%
  5. #46low of 570.2%
  6. #55none of 570.0%

Leader: GPT-6 Astra · 99.9%

reasoning

BullshitBench v2

196 results
  1. #53none of 19659.0%
  2. #55max of 19658.0%

Leader: Opus 4.8 · 95.0%

math

FrontierMath Tier 4 (v2)

70 results
  1. #21max of 7056.1%

Leader: GPT-6.1 Sol · 100.0%

agentic

Terminal-Bench 4.0

35 results
  1. #31max of 3516.4%

Leader: Opus 5.5 · 64.8%

All 75 benchmarks

show 69 more

agentic

AA-AnalystAgent

44 results
  1. #17max of 4443.8%

Leader: Gemini 3.7 Flash · 60.0%

agentic

AA-Briefcase v1.1

224 results
  1. #64max of 2241336
  2. #83xhigh of 2241237
  3. #89high of 2241186
  4. #102medium of 2241099
  5. #135none of 224903
  6. #154low of 224741

Leader: Sonnet 5.5 · 1823

agentic

Agent Arena

50 results
  1. #26max of 501.9%

Leader: Opus 5.5 · 15.5%

agentic

AutomationBench (Zapier)

115 results
  1. #36max of 11520.7%
  2. #55high of 11514.5%
  3. #63xhigh of 11512.6%
  4. #74medium of 1159.4%
  5. #109low of 1151.2%
  6. #113none of 1150.3%

Leader: Gemini 4 · 51.3%

agentic

AutomationBench-AA

219 results
  1. #66max of 21953.2%
  2. #79high of 21947.8%
  3. #81xhigh of 21947.8%
  4. #93medium of 21940.5%
  5. #145low of 21912.0%
  6. #148none of 2199.1%

Leader: Gemini 4 · 77.5%

agentic

BioMysteryBench

27 results
  1. #24max of 2761.5%

Leader: Sonnet 5.5 · 81.1%

agentic

Harvey's Legal Agent Benchmark

79 results
  1. #43max of 792.9%

Leader: Muse Spark 1.2 · 25.4%

agentic

Legal Research Bench

78 results
  1. #46max of 7830.3%

Leader: Muse Spark 1.3 · 55.3%

agentic

Tax Agent Bench

69 results
  1. #46max of 6919.9%

Leader: Fable 5.1 · 49.2%

agentic

Terminal-Bench 4.0 (Vals)

47 results
  1. #31max of 4713.6%

Leader: Opus 5.5 · 65.2%

agentic

Terminal-Bench Science (Vals)

40 results
  1. #25max of 404.3%

Leader: GPT-6 Astra · 62.9%

agentic

WeirdML v3

17 results
  1. #11xhigh of 177.9%

Leader: GPT-6 Astra · 42.2%

coding

Arena Image-to-WebDev

58 results
  1. #24max of 581570

Leader: Opus 5.5 · 1749

coding

Arena WebDev

122 results
  1. #33max of 1221581

Leader: Opus 5.5 · 1813

coding

Code Migration

77 results
  1. #27max of 7742.6%

Leader: Sonnet 5.5 · 69.8%

coding

FrontierCode 1.1 Extended

130 results
  1. #56max of 13056.1%
  2. #79xhigh of 13051.4%
  3. #82high of 13051.3%
  4. #90medium of 13050.1%
  5. #112low of 13039.1%

Leader: Opus 5.5 · 65.3%

coding

FrontierCode 1.1 Main

130 results
  1. #56max of 13042.4%
  2. #80high of 13037.3%
  3. #81xhigh of 13037.1%
  4. #87medium of 13035.5%
  5. #108low of 13025.7%

Leader: Opus 5.5 · 54.6%

coding

IOI

44 results
  1. #21max of 4455.6%

Leader: Gemini 4 · 100.0%

coding

ProgramBench v1 Almost Resolved

62 results
  1. #33max of 628.0%

Leader: Opus 5.5 · 65.0%

coding

ProgramBench v1 Fully Resolved

62 results
  1. #25max of 620.5%

Leader: Opus 5.5 · 18.5%

coding

ProgramBench v1 Raw Pass Rate

62 results
  1. #29max of 6266.2%

Leader: Opus 5.5 · 87.0%

coding

SciCode

226 results
  1. #58max of 22654.6%
  2. #88xhigh of 22651.7%
  3. #97medium of 22650.9%
  4. #104high of 22650.3%
  5. #130low of 22646.9%
  6. #151none of 22643.1%

Leader: Opus 5.5 · 66.9%

coding

Terminal-Bench 2.1 (Vals)

74 results
  1. #23max of 7473.0%

Leader: Opus 5.5 · 87.6%

coding

Terminal-Bench Science-AA

46 results
  1. #29max of 468.6%

Leader: GPT-6 Astra · 63.3%

coding

Terminal-Bench v4-AA

216 results
  1. #77max of 21612.6%
  2. #90xhigh of 2168.1%
  3. #101high of 2164.5%
  4. #109medium of 2162.5%
  5. #122none of 2161.5%
  6. #158low of 2160.0%

Leader: Sonnet 5.5 · 63.6%

coding

Vibe Code Bench v1.1

105 results
  1. #22max of 10581.6%

Leader: Sonnet 5.5 · 92.4%

composite

AA Intelligence Index v4.3.2

610 results
  1. #75max of 61038.1
  2. #86xhigh of 61034.6
  3. #104high of 61032.9
  4. #121medium of 61029.9
  5. #206low of 61021.5
  6. #246none of 61018.5

Leader: Opus 5.5 · 57.6

composite

Arena Text

220 results
  1. #76max of 2201443

Leader: Gemini 4 · 1525

composite

Arena Text — Multi-Turn

220 results
  1. #85max of 2201444

Leader: Gemini 4 · 1553

composite

Vals Index

47 results
  1. #28max of 4751.2%

Leader: Gemini 4 · 68.9%

cost

AA Intelligence Index Cost per Task

177 results
  1. #1low of 177$0.00
  2. #7none of 177$0.01
  3. #12medium of 177$0.02
  4. #18high of 177$0.03
  5. #20xhigh of 177$0.04
  6. #30max of 177$0.07

Leader: this model · $0.00

cost

Cost to Run AA Intelligence Index

177 results
  1. #1low of 177$10.63
  2. #4none of 177$19.16
  3. #10medium of 177$31.17
  4. #17high of 177$47.85
  5. #22xhigh of 177$66.81
  6. #30max of 177$122

Leader: this model · $10.63

creative

threejseval

34 results
  1. #24high of 341386

Leader: Opus 5.5 · 2079

creative

VoxelBench (text)

55 results
  1. #16max of 551800

Leader: GPT-6 Astra · 2732

economics

Excel Modeling Benchmark

74 results
  1. #15max of 7468.5%

Leader: Fable 5.1 · 76.7%

economics

Finance Agent v2

78 results
  1. #43max of 7849.9%

Leader: Gemini 4 · 65.4%

economics

GDP.pdf

210 results
  1. #47max of 21022.8%
  2. #70xhigh of 21018.6%
  3. #74high of 21018.2%
  4. #99medium of 21013.8%
  5. #135low of 2109.0%
  6. #153none of 2105.8%

Leader: Muse Wonder Full (Max) · 32.6%

economics

GDPval-AA v2.1

288 results
  1. #65max of 2881430
  2. #80xhigh of 2881364
  3. #89high of 2881340
  4. #103medium of 2881262
  5. #148none of 2881045
  6. #153low of 2881033

Leader: Opus 5.5 · 1866

knowledge

AA-Omniscience Accuracy

526 results
  1. #93xhigh of 52644.2%
  2. #94max of 52643.8%
  3. #99medium of 52643.1%
  4. #103high of 52642.8%
  5. #117low of 52640.8%
  6. #167none of 52631.9%

Leader: Fable 5.1 · 67.2%

knowledge

AA-Omniscience Non-hallucination

526 results
  1. #251max of 52623.3%
  2. #267none of 52621.1%
  3. #307xhigh of 52617.6%
  4. #323low of 52615.7%
  5. #324high of 52615.6%
  6. #328medium of 52615.3%

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

knowledge

Harvey LAB-AA v1.1

27 results
  1. #11max of 273.3%

Leader: Muse Wonder Full (Max) · 13.1%

knowledge

MedCode

94 results
  1. #38max of 9444.7%

Leader: Opus 5 · 63.6%

knowledge

MedScribe

96 results
  1. #40max of 9683.7%

Leader: Opus 5.5 · 91.4%

knowledge

MMMU-Pro

265 results
  1. #55xhigh of 26579.7%
  2. #56max of 26579.7%
  3. #76medium of 26577.9%
  4. #78high of 26577.5%
  5. #90low of 26576.0%
  6. #189none of 26562.1%

Leader: Opus 5.5 · 87.7%

knowledge

Public Benefits Bench v1.1

50 results
  1. #38max of 5057.6%

Leader: Opus 5 · 76.9%

long-context

AA-LCR v1.1

523 results
  1. #21max of 52383.3%
  2. #80xhigh of 52380.0%
  3. #97high of 52379.3%
  4. #112medium of 52378.3%
  5. #170low of 52374.0%
  6. #344none of 52339.7%

Leader: Kimi K3 · 88.7%

long-context

MLCR-AA

94 results
  1. #43max of 9416.1%

Leader: Sonnet 5.5 · 75.0%

math

FrontierMath Tiers 1–3 (v2)

115 results
  1. #16max of 11578.9%

Leader: GPT-6.1 Sol · 93.7%

math

ProofBench v1.1

52 results
  1. #21max of 5264.0%

Leader: Sonnet 5.5 · 100.0%

other

Arena Vision

116 results
  1. #61max of 1161241

Leader: Fable 5 · 1308

other

T3 Code usage share

85 results
  1. #10max of 851.2%

Leader: Opus 5.5 · 29.1%

reasoning

ARC-AGI-1

215 results
  1. #88max of 21586.7%
  2. #118xhigh of 21573.0%
  3. #123high of 21570.3%
  4. #137medium of 21561.0%
  5. #168low of 21537.7%
  6. #197none of 2158.8%

Leader: Fable 5 · 98.5%

reasoning

ARC-AGI-2

216 results
  1. #82max of 21659.3%
  2. #103xhigh of 21641.9%
  3. #115high of 21631.4%
  4. #127medium of 21618.1%
  5. #158low of 2164.6%
  6. #216none of 2160.0%

Leader: GPT-6 Astra · 95.0%

reasoning

Blueprint-Bench 2

28 results
  1. #16max of 2831.2%

Leader: Gemini 4 · 54.4%

reasoning

CritPt

529 results
  1. #62max of 52919.4%
  2. #76xhigh of 52917.4%
  3. #88high of 52915.4%
  4. #113medium of 52910.6%
  5. #185low of 5292.6%
  6. #223none of 5291.1%

Leader: Muse Wonder Full (Max) · 32.9%

reasoning

Furniture Assembly

31 results
  1. #11max of 3144.2%

Leader: Opus 5.5 · 83.3%

reasoning

Humanity's Sixth Sense

25 results
  1. #25max of 2521.0%

Leader: GPT-6 Astra · 53.6%

reasoning

MysteryMechanism

27 results
  1. #23max of 2719.4%

Leader: GPT-6 Astra · 53.2%

reasoning

Roboflow Visual Reasoning

102 results
  1. #38high of 10271.1%
  2. #52low of 10264.2%

Leader: GPT-6 Astra · 91.2%

reasoning

SAGE

80 results
  1. #28max of 8048.1%

Leader: Opus 4.7 · 56.1%

reliability

Tool-call error rate

171 results
  1. #41max of 1710.4%

Leader: GPT-5.4 Pro · 0.0%

reliability

Uptime (7d)

181 results
  1. #169max of 18196.3%

Leader: GPT-5 Pro · 100.0%

safety

CWE-Bench

19 results
  1. #8max of 1956.7%

Leader: Grok 4.7 · 68.3%

safety

CyberBench v1.1

45 results
  1. #5max of 4576.3%

Leader: Wonder · 80.4%

safety

SRE Bench

32 results
  1. #21max of 322.7%

Leader: GPT-6 Astra · 56.9%

speed

Latency (Time to First Token)

351 results
  1. #59none of 3510.9s
  2. #186low of 3512.5s
  3. #278high of 35120s
  4. #312xhigh of 35146s
  5. #328max of 351101s

Leader: Gemini 2.5 Flash-Lite · 0.3s

speed

Output Speed (tokens/s)

351 results
  1. #104max of 351134.6
  2. #105high of 351133.6
  3. #109xhigh of 351128.3
  4. #113low of 351126.0
  5. #134none of 351112.3

Leader: Celeris-1 · 1522.5

speed

Provider latency

177 results
  1. #134max of 1772.4s

Leader: Qwen 2.5 7B · 0.2s

speed

Provider throughput

176 results
  1. #73max of 17673.0

Leader: Muse Spark 1.1 · 204.3