priorsBuilder

Anthropic · released 2026-09-28

Sonnet 5.5

Measured on 74 benchmarks across 5 reasoning efforts.

General frontier

67.9

#6 of 32 models · best at max

Reasoning efforts in your mix

max67.9#18 of 128
xhigh62.2#30 of 128
high58.5#34 of 128
medium49.7#50 of 128
low47.1#57 of 128

Your mix’s benchmarks

6 of 8 measured

reasoning

Humanity's Last Exam

584 results
  1. #9max of 58455.0%
  2. #23xhigh of 58450.0%
  3. #44high of 58445.8%
  4. #82medium of 58439.8%
  5. #107low of 58436.2%

Leader: Opus 5.5 · 61.4%

knowledge

AA-Omniscience

523 results
  1. #26max of 52332.3
  2. #52xhigh of 52323.5
  3. #60high of 52320.9
  4. #65medium of 52320.1
  5. #68low of 52319.4

Leader: Opus 5.5 · 46.4

reasoning

BullshitBench v2

192 results
  1. #45low of 19260.0%
  2. #52max of 19258.0%

Leader: Opus 4.8 · 95.0%

math

FrontierMath Tier 4 (v2)

70 results
  1. #13max of 7080.5%

Leader: GPT-6.1 Sol · 100.0%

reasoning

SimpleBench

97 results
  1. #15max of 9775.9%

Leader: Opus 5.5 · 88.4%

agentic

Terminal-Bench 4.0

35 results
  1. #2max of 3561.8%

Leader: Opus 5.5 · 64.8%

All 74 benchmarks

show 68 more

agentic

AA-AnalystAgent

39 results
  1. #2max of 3957.5%

Leader: Gemini 3.7 Flash · 60.0%

agentic

AA-Briefcase v1.1

222 results
  1. #1max of 2221823
  2. #4xhigh of 2221752
  3. #10high of 2221638
  4. #45medium of 2221442
  5. #73low of 2221271

Leader: this model · 1823

agentic

Agent Arena

49 results
  1. #4max of 4912.0%

Leader: Opus 5.5 · 14.3%

agentic

AutomationBench (Zapier)

115 results
  1. #3max of 11544.8%
  2. #8xhigh of 11536.8%
  3. #20high of 11529.4%
  4. #31medium of 11523.3%
  5. #41low of 11518.3%

Leader: Gemini 4 · 51.3%

agentic

AutomationBench-AA

217 results
  1. #2max of 21771.8%
  2. #12xhigh of 21765.5%
  3. #35high of 21759.4%
  4. #55medium of 21754.9%
  5. #74low of 21749.4%

Leader: Gemini 4 · 77.5%

agentic

BioMysteryBench

25 results
  1. #1max of 2581.1%

Leader: this model · 81.1%

agentic

Harvey's Legal Agent Benchmark

76 results
  1. #41max of 762.9%

Leader: Muse Spark 1.2 · 25.4%

agentic

Legal Research Bench

75 results
  1. #10max of 7548.1%

Leader: Muse Spark 1.3 · 55.3%

agentic

Tax Agent Bench

67 results
  1. #6max of 6742.0%

Leader: Fable 5.1 · 49.2%

agentic

Terminal-Bench 4.0 (Vals)

45 results
  1. #2max of 4564.1%

Leader: Opus 5.5 · 65.2%

agentic

Terminal-Bench Science (Vals)

39 results
  1. #4max of 3945.7%

Leader: GPT-6 Astra · 62.9%

agentic

Vending-Bench 2

67 results
  1. #5max of 67$11,165

Leader: GPT-6 Astra · $15,515

agentic

WeirdML v3

17 results
  1. #5xhigh of 1719.9%

Leader: GPT-6 Astra · 42.2%

coding

Arena Image-to-WebDev

58 results
  1. #2xhigh of 581740

Leader: Opus 5.5 · 1749

coding

Arena WebDev

122 results
  1. #3xhigh of 1221774
  2. #6high of 1221715

Leader: Opus 5.5 · 1813

coding

Code Migration

75 results
  1. #1max of 7569.8%

Leader: this model · 69.8%

coding

FrontierCode 1.1 Extended

130 results
  1. #5xhigh of 13064.4%
  2. #19high of 13061.5%
  3. #39max of 13059.1%
  4. #89medium of 13050.5%
  5. #105low of 13043.4%

Leader: Opus 5.5 · 65.3%

coding

FrontierCode 1.1 Main

130 results
  1. #8xhigh of 13052.1%
  2. #20high of 13049.4%
  3. #40max of 13046.2%
  4. #85medium of 13036.5%
  5. #100low of 13029.3%

Leader: Opus 5.5 · 54.6%

coding

FrontierSWE v2

21 results
  1. #3max of 2161.9%

Leader: GPT-6 Astra · 65.5%

coding

IOI

42 results
  1. #9max of 4283.1%

Leader: Gemini 4 · 100.0%

coding

ProgramBench v1 Almost Resolved

62 results
  1. #4max of 6249.5%

Leader: Opus 5.5 · 65.0%

coding

ProgramBench v1 Fully Resolved

62 results
  1. #3max of 626.5%

Leader: Opus 5.5 · 18.5%

coding

ProgramBench v1 Raw Pass Rate

62 results
  1. #10max of 6276.3%

Leader: Opus 5.5 · 87.0%

coding

SciCode

224 results
  1. #5max of 22461.0%
  2. #27xhigh of 22457.3%
  3. #70high of 22453.7%
  4. #77medium of 22452.9%
  5. #114low of 22449.1%

Leader: Opus 5.5 · 66.9%

coding

Terminal-Bench 2.1 (Vals)

74 results
  1. #7high of 7483.1%

Leader: Opus 5.5 · 87.6%

coding

Terminal-Bench Science-AA

45 results
  1. #5max of 4553.3%
  2. #6xhigh of 4552.4%
  3. #10high of 4534.3%
  4. #19medium of 4516.2%
  5. #23low of 4510.5%

Leader: GPT-6 Astra · 63.3%

coding

Terminal-Bench v4-AA

214 results
  1. #1max of 21463.6%
  2. #7xhigh of 21457.1%
  3. #24high of 21443.9%
  4. #43medium of 21429.8%
  5. #58low of 21420.7%

Leader: this model · 63.6%

coding

Vibe Code Bench v1.1

103 results
  1. #1max of 10392.4%

Leader: this model · 92.4%

composite

AA Intelligence Index v4.3.2

607 results
  1. #2max of 60756.0
  2. #10xhigh of 60751.9
  3. #28high of 60746.8
  4. #56medium of 60740.8
  5. #81low of 60735.9

Leader: Opus 5.5 · 57.6

composite

Arena Text

220 results
  1. #31xhigh of 2201476

Leader: Gemini 4 · 1525

composite

Arena Text — Multi-Turn

220 results
  1. #29xhigh of 2201484

Leader: Gemini 4 · 1553

composite

Vals Index

45 results
  1. #2max of 4567.0%

Leader: Gemini 4 · 68.9%

cost

AA Intelligence Index Cost per Task

177 results
  1. #77low of 177$0.35
  2. #88medium of 177$0.48
  3. #109high of 177$0.88
  4. #150xhigh of 177$2.01
  5. #172max of 177$5.46

Leader: GPT-6 Luna · $0.00

cost

Cost to Run AA Intelligence Index

177 results
  1. #71low of 177$482
  2. #80medium of 177$622
  3. #98high of 177$1,027
  4. #141xhigh of 177$2,177
  5. #172max of 177$7,260

Leader: GPT-6 Luna · $10.63

creative

MineBench

72 results
  1. #6max of 722039

Leader: GPT-6 Astra Pro · 2278

creative

VoxelBench (text)

54 results
  1. #54max of 541000

Leader: GPT-6 Astra · 2654

economics

Excel Modeling Benchmark

72 results
  1. #3max of 7275.7%

Leader: Fable 5.1 · 76.7%

economics

Finance Agent v2

76 results
  1. #10max of 7658.1%

Leader: Gemini 4 · 65.4%

economics

GDP.pdf

208 results
  1. #24max of 20825.8%
  2. #28high of 20825.2%
  3. #29xhigh of 20824.6%
  4. #59medium of 20820.2%
  5. #85low of 20816.0%

Leader: GPT-6 Astra · 32.2%

economics

GDPval-AA v2.1

286 results
  1. #2max of 2861838
  2. #6xhigh of 2861730
  3. #39high of 2861550
  4. #92medium of 2861323
  5. #118low of 2861175

Leader: Opus 5.5 · 1866

knowledge

AA-Omniscience Accuracy

523 results
  1. #42max of 52353.9%
  2. #48xhigh of 52353.0%
  3. #53high of 52352.0%
  4. #75medium of 52347.2%
  5. #80low of 52346.3%

Leader: Fable 5.1 · 67.2%

knowledge

AA-Omniscience Non-hallucination

523 results
  1. #100max of 52353.0%
  2. #110low of 52349.8%
  3. #116medium of 52348.8%
  4. #170xhigh of 52337.1%
  5. #180high of 52335.4%

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

knowledge

Harvey LAB-AA v1.1

25 results
  1. #11max of 252.8%

Leader: Grok 4.7 · 9.4%

knowledge

MedCode

94 results
  1. #12max of 9452.9%

Leader: Opus 5 · 63.6%

knowledge

MedScribe

96 results
  1. #3max of 9691.1%

Leader: Opus 5.5 · 91.4%

knowledge

Public Benefits Bench v1.1

48 results
  1. #13max of 4867.2%

Leader: Opus 5 · 76.9%

long-context

AA-LCR v1.1

520 results
  1. #33max of 52082.7%
  2. #84xhigh of 52079.7%
  3. #115high of 52078.0%
  4. #139medium of 52076.3%
  5. #142low of 52076.0%

Leader: Kimi K3 · 88.7%

long-context

MLCR-AA

93 results
  1. #1max of 9375.0%

Leader: this model · 75.0%

math

FrontierMath Erdős

7 results
  1. #3max of 72.9%

Leader: GPT-6 Astra · 2.9%

math

FrontierMath Tiers 1–3 (v2)

115 results
  1. #7max of 11588.8%

Leader: GPT-6.1 Sol · 93.7%

math

ProofBench v1.1

50 results
  1. #1max of 50100.0%

Leader: this model · 100.0%

other

Arena Vision

116 results
  1. #38xhigh of 1161266

Leader: Fable 5 · 1308

other

T3 Code usage share

85 results
  1. #11max of 851.1%

Leader: Opus 5.5 · 29.1%

reasoning

Blueprint-Bench 2

28 results
  1. #5max of 2839.3%

Leader: Gemini 4 · 54.4%

reasoning

CritPt

526 results
  1. #7max of 52631.4%
  2. #10xhigh of 52631.1%
  3. #44high of 52624.6%
  4. #77medium of 52616.9%
  5. #104low of 52611.4%

Leader: GPT-5.6 Sol · 32.3%

reasoning

Furniture Assembly

31 results
  1. #4max of 3175.0%

Leader: Opus 5.5 · 83.3%

reasoning

MysteryMechanism

25 results
  1. #3max of 2549.1%

Leader: GPT-6 Astra · 53.2%

reasoning

Roboflow Visual Reasoning

102 results
  1. #6high of 10283.9%
  2. #19low of 10276.4%

Leader: GPT-6 Astra · 91.2%

reasoning

SAGE

80 results
  1. #11max of 8051.8%

Leader: Opus 4.7 · 56.1%

reasoning

SimpleQA Verified

47 results
  1. #18max of 4744.7%

Leader: Gemini 3.1 Pro · 77.5%

reliability

Tool-call error rate

171 results
  1. #98max of 1712.0%

Leader: GPT-5.4 Pro · 0.0%

reliability

Uptime (7d)

181 results
  1. #47max of 18199.9%

Leader: GPT-5 Pro · 100.0%

safety

CyberBench v1.1

43 results
  1. #33max of 4359.6%

Leader: GPT-6 Sol · 78.0%

safety

SRE Bench

32 results
  1. #8max of 3230.2%

Leader: GPT-6 Astra · 56.9%

speed

Latency (Time to First Token)

355 results
  1. #34low of 3550.9s
  2. #135medium of 3551.9s
  3. #266high of 35512s
  4. #307xhigh of 35533s
  5. #354max of 355497s

Leader: Gemini 2.5 Flash-Lite · 0.3s

speed

Output Speed (tokens/s)

355 results
  1. #89max of 355141.8
  2. #132medium of 355108.7
  3. #133high of 355108.7
  4. #134xhigh of 355108.5
  5. #137low of 355108.2

Leader: Celeris-1 · 1461.1

speed

Provider latency

178 results
  1. #97max of 1781.4s

Leader: Qwen 2.5 7B · 0.2s

speed

Provider throughput

177 results
  1. #24max of 177103.5

Leader: Llama 3.2 3b · 199.5