- #65max of 2241334
- #66xhigh of 2241333
- #87high of 2241202
- #113medium of 2241039
- #115low of 2241014
- #125none of 224947
Leader: Sonnet 5.5 · 1823
- #31xhigh of 500.9%
Leader: Opus 5.5 · 15.5%
- #4max of 3338.9%
Leader: Gemini 3.5 Flash · 47.1%
agentic
AutomationBench (Zapier)
115 results- #30max of 11523.6%
- #54xhigh of 11514.8%
- #66high of 11511.7%
- #81medium of 1158.1%
- #84low of 1157.5%
Leader: Gemini 4 · 51.3%
agentic
AutomationBench-AA
219 results- #36max of 21959.6%
- #84xhigh of 21947.1%
- #90high of 21942.0%
- #109medium of 21933.5%
- #112low of 21929.1%
- #143none of 21912.5%
Leader: Gemini 4 · 77.5%
agentic
EnterpriseOps-Gym-AA
50 results- #32max of 5038.5%
Leader: Fable 5 · 51.1%
agentic
Harvey's Legal Agent Benchmark
79 results- #59max of 790.8%
Leader: Muse Spark 1.2 · 25.4%
- #5max of 4551.0%
Leader: GPT-5.6 Sol · 56.2%
agentic
Legal Research Bench
78 results- #26max of 7841.3%
Leader: Muse Spark 1.3 · 55.3%
- #10max of 3458.9%
Leader: DeepSeek V4.1 Flash · 69.8%
- #26max of 6930.0%
Leader: Fable 5.1 · 49.2%
agentic
Terminal-Bench 2.1
18 results- #11max of 1878.4%
Leader: GPT-6 Astra · 87.4%
agentic
Terminal-Bench 4.0 (Vals)
47 results- #22max of 4722.7%
Leader: Opus 5.5 · 65.2%
agentic
Terminal-Bench Science (Vals)
40 results- #17max of 4010.0%
Leader: GPT-6 Astra · 62.9%
- #15max of 67$7,343
Leader: GPT-6 Astra · $15,515
- #88max of 39386.3%
- #124xhigh of 39380.4%
- #132high of 39378.4%
- #149medium of 39372.8%
- #182low of 39360.5%
Leader: GLM 5.2 · 99.1%
- #29max of 20640.2%
- #68xhigh of 20629.7%
- #73high of 20628.7%
- #80medium of 20625.6%
- #103low of 20618.8%
- #116none of 20615.7%
Leader: Qwen 3.8 Max · 51.3%
coding
Arena Image-to-WebDev
58 results- #29xhigh of 581540
Leader: Opus 5.5 · 1749
- #47xhigh of 1221522
Leader: Opus 5.5 · 1813
- #13max of 7747.8%
Leader: Sonnet 5.5 · 69.8%
coding
FrontierCode 1.1 Extended
130 results- #59max of 13055.8%
- #76xhigh of 13052.6%
- #85high of 13050.9%
- #95medium of 13047.7%
- #114low of 13037.8%
Leader: Opus 5.5 · 65.3%
coding
FrontierCode 1.1 Main
130 results- #64max of 13041.3%
- #75xhigh of 13038.8%
- #83high of 13036.9%
- #95medium of 13033.9%
- #112low of 13024.1%
Leader: Opus 5.5 · 54.6%
- #7max of 4487.6%
Leader: Gemini 4 · 100.0%
coding
LiveCodeBench (Vals)
127 results- #27xhigh of 12785.9%
Leader: Fable 5.1 · 90.5%
coding
ProgramBench v1 Almost Resolved
62 results- #19max of 6215.0%
Leader: Opus 5.5 · 65.0%
coding
ProgramBench v1 Fully Resolved
62 results- #24max of 620.5%
Leader: Opus 5.5 · 18.5%
coding
ProgramBench v1 Raw Pass Rate
62 results- #13max of 6272.3%
Leader: Opus 5.5 · 87.0%
- #53max of 22655.0%
- #82high of 22652.4%
- #83xhigh of 22652.3%
- #102medium of 22650.5%
- #109low of 22649.9%
- #140none of 22645.1%
Leader: Opus 5.5 · 66.9%
coding
SWE-bench Verified (Vals)
85 results- #5max of 8595.4%
Leader: Opus 5 · 97.0%
coding
Terminal-Bench 2.1 (Vals)
74 results- #15max of 7477.5%
Leader: Opus 5.5 · 87.6%
coding
Terminal-Bench Hard-AA
387 results- #4xhigh of 38762.9%
- #11max of 38757.6%
- #13high of 38757.6%
- #38low of 38743.9%
Leader: GPT-5.6 Sol · 65.9%
coding
Terminal-Bench Science-AA
46 results- #20max of 4613.8%
Leader: GPT-6 Astra · 63.3%
coding
Terminal-Bench v2.1-AA
238 results- #17max of 23888.0%
- #45xhigh of 23880.1%
- #60high of 23875.7%
- #68medium of 23872.3%
- #89low of 23862.5%
- #105none of 23856.2%
Leader: Fable 5.1 · 91.4%
coding
Terminal-Bench v4-AA
216 results- #33max of 21635.4%
- #86xhigh of 21610.1%
- #117high of 2161.5%
- #120low of 2161.5%
- #125medium of 2161.0%
- #140none of 2160.5%
Leader: Sonnet 5.5 · 63.6%
coding
Vibe Code Bench 1-100
22 results- #15max of 2214.8%
Leader: Opus 5.5 · 30.4%
coding
Vibe Code Bench v1.1
105 results- #34max of 10574.6%
Leader: Sonnet 5.5 · 92.4%
- #18max of 12778.3%
Leader: GPT-6 Astra Pro · 93.6%
composite
AA Intelligence Index v4.3.2
610 results- #52max of 61042.1
- #76xhigh of 61038.0
- #92high of 61034.2
- #118medium of 61030.1
- #139low of 61027.5
- #216none of 61020.8
Leader: Opus 5.5 · 57.6
- #45xhigh of 2201466
Leader: Gemini 4 · 1525
composite
Arena Text — Multi-Turn
220 results- #53xhigh of 2201469
Leader: Gemini 4 · 1553
- #22max of 4753.1%
Leader: Gemini 4 · 68.9%
composite
Vals Multimodal Index v1.2
33 results- #14xhigh of 3365.1%
Leader: Fable 5 · 74.2%
cost
AA Intelligence Index Cost per Task
177 results- #48none of 177$0.14
- #49low of 177$0.14
- #55medium of 177$0.18
- #75high of 177$0.34
- #98xhigh of 177$0.63
- #133max of 177$1.40
Leader: GPT-6 Luna · $0.00
cost
Cost to Run AA Intelligence Index
177 results- #48none of 177$256
- #56low of 177$350
- #64medium of 177$447
- #85high of 177$768
- #106xhigh of 177$1,187
- #145max of 177$2,501
Leader: GPT-6 Luna · $10.63
- #20max of 341449
- #30low of 341172
Leader: Opus 5.5 · 2079
creative
VoxelBench (text)
55 results- #12max of 551887
Leader: GPT-6 Astra · 2732
- #30xhigh of 11765.3%
Leader: Opus 5 · 73.2%
economics
Excel Modeling Benchmark
74 results- #22max of 7466.2%
Leader: Fable 5.1 · 76.7%
economics
Finance Agent v2
78 results- #25max of 7854.4%
Leader: Gemini 4 · 65.4%
- #33xhigh of 21024.6%
- #38max of 21024.0%
- #59high of 21020.8%
- #77low of 21017.6%
- #84medium of 21017.0%
- #121none of 21011.2%
Leader: Muse Wonder Full (Max) · 32.6%
- #59max of 2881451
- #62xhigh of 2881447
- #78high of 2881376
- #102medium of 2881264
- #134low of 2881108
- #140none of 2881097
Leader: Opus 5.5 · 1866
economics
MortgageTax (Vals)
84 results- #25xhigh of 8467.3%
Leader: Opus 5 · 72.1%
- #6xhigh of 12576.2%
Leader: Muse Spark 1.2 · 80.4%
- #57max of 39971.2%
- #96xhigh of 39966.3%
- #107high of 39964.4%
- #118medium of 39962.2%
- #126low of 39959.7%
Leader: Grok 4.3 · 83.3%
knowledge
AA-Omniscience Accuracy
526 results- #77max of 52646.8%
- #86xhigh of 52645.5%
- #87high of 52645.5%
- #91medium of 52644.6%
- #95low of 52643.7%
- #144none of 52636.8%
Leader: Fable 5.1 · 67.2%
knowledge
AA-Omniscience Non-hallucination
526 results- #368max of 52612.1%
- #384xhigh of 52611.0%
- #396medium of 52610.2%
- #397high of 52610.2%
- #400low of 52610.1%
- #499none of 5265.0%
Leader: MiniCPM5-1B (Non-reasoning) · 99.1%
- #35max of 53992.5%
- #55xhigh of 53990.8%
- #71high of 53989.6%
- #101medium of 53987.2%
- #144low of 53984.3%
- #254none of 53974.6%
Leader: GPT-6 Astra · 96.3%
knowledge
GPQA Diamond (Vals)
122 results- #24xhigh of 12290.9%
Leader: Gemini 3.1 Pro · 95.5%
- #42xhigh of 9443.4%
Leader: Opus 5 · 63.6%
- #46xhigh of 9682.9%
Leader: Opus 5.5 · 91.4%
- #39xhigh of 12286.7%
Leader: Fable 5.1 · 92.4%
- #20max of 8186.5%
Leader: Fable 5.1 · 90.6%
- #42max of 26580.7%
- #57xhigh of 26579.5%
- #60high of 26579.1%
- #82medium of 26576.8%
- #89low of 26576.1%
- #164none of 26566.7%
Leader: Opus 5.5 · 87.7%
knowledge
Public Benefits Bench v1.1
50 results- #28max of 5062.4%
Leader: Opus 5 · 76.9%
- #29max of 52383.0%
- #104xhigh of 52379.0%
- #124high of 52377.7%
- #166medium of 52374.0%
- #200low of 52371.3%
- #277none of 52358.7%
Leader: Kimi K3 · 88.7%
long-context
Arena Document
43 results- #13xhigh of 431472
Leader: Opus 5 · 1516
- #19max of 9431.7%
- #26xhigh of 9421.7%
- #57high of 9412.2%
- #59medium of 9412.2%
- #68low of 948.9%
Leader: Sonnet 5.5 · 75.0%
math
FrontierMath Tiers 1–3 (v2)
115 results- #10max of 11586.0%
Leader: GPT-6.1 Sol · 93.7%
- #19max of 5274.0%
Leader: Sonnet 5.5 · 100.0%
- #37xhigh of 1161268
Leader: Fable 5 · 1308
- #12max of 850.7%
Leader: Opus 5.5 · 29.1%
- #23max of 21596.5%
- #41xhigh of 21594.0%
- #56high of 21592.0%
- #111medium of 21577.0%
- #139low of 21560.2%
Leader: Fable 5 · 98.5%
- #33max of 21683.9%
- #45xhigh of 21674.2%
- #62high of 21667.1%
- #107medium of 21637.5%
- #125low of 21618.8%
Leader: GPT-6 Astra · 95.0%
reasoning
Blueprint-Bench 2
28 results- #17max of 2830.8%
Leader: Gemini 4 · 54.4%
- #17max of 52930.0%
- #35xhigh of 52927.1%
- #52high of 52922.9%
- #77medium of 52917.4%
- #118low of 5299.4%
- #197none of 5292.0%
Leader: Muse Wonder Full (Max) · 32.9%
reasoning
Furniture Assembly
31 results- #9max of 3154.2%
Leader: Opus 5.5 · 83.3%
reasoning
Humanity's Sixth Sense
25 results- #23max of 2522.7%
Leader: GPT-6 Astra · 53.6%
- #24max of 13585.1%
Leader: Fable 5 · 88.6%
reasoning
Roboflow Visual Reasoning
102 results- #50high of 10265.3%
- #61low of 10260.9%
Leader: GPT-6 Astra · 91.2%
- #33max of 8047.0%
Leader: Opus 4.7 · 56.1%
reasoning
SimpleQA Verified
47 results- #20max of 4741.5%
Leader: Gemini 3.1 Pro · 77.5%
reliability
Tool-call error rate
171 results- #26max of 1710.3%
Leader: GPT-5.4 Pro · 0.0%
- #93max of 18199.5%
Leader: GPT-5 Pro · 100.0%
- #18max of 4572.1%
Leader: Wonder · 80.4%
- #12max of 329.5%
Leader: GPT-6 Astra · 56.9%
speed
Latency (Time to First Token)
351 results- #23none of 3510.7s
- #127low of 3511.7s
- #137medium of 3511.7s
- #242high of 3517.4s
- #309xhigh of 35143s
- #338max of 351129s
Leader: Gemini 2.5 Flash-Lite · 0.3s
speed
Output Speed (tokens/s)
351 results- #107max of 351129.3
- #140xhigh of 351110.4
- #152medium of 351105.4
- #153high of 351105.0
- #157none of 351102.4
- #161low of 351101.2
Leader: Celeris-1 · 1522.5
- #128max of 1772.2s
Leader: Qwen 2.5 7B · 0.2s
- #61max of 17679.0
Leader: Muse Spark 1.1 · 204.3