AIEvals

Models

Claude Opus 5

The preceding Claude Opus release, retained for comparison with Opus 5.5.

AnthropicReleased 22 July 2026Context 1M tokensOutput 128k tokensPrivate weights$5.00 in / $25.00 out per Mtok

Family Claude Opus 566 benchmarks with a published result6 independent first places 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 AI63.67%5of 335of 41Gemini 4 Argon5.23 pts
Artificial Analysis Intelligence IndexArtificial Analysis517of 42no list readClaude Opus 5.57 pts
Terminal-Bench 4.0Artificial Analysis48.99%7of 40no list readClaude Sonnet 5.514.65 pts
SWE-bench (Verified)Vals AI97.00%1of 231of 88Claude Opus 5—
LiveCodeBenchVals AI89.03%4of 214of 143Claude Fable 5.11.49 pts
GPQA DiamondVals AI93.43%7of 237of 138Gemini 3.1 Pro2.02 pts
MMLU ProVals AI91.59%2of 242of 138Claude Fable 5.10.79 pts
MMMU ProVals AI89.88%2of 202of 93Claude Fable 5.10.76 pts
Vibe Code Bench v1.1Vals AI88.40%8of 388of 106Claude Sonnet 5.53.99 pts
Terminal-Bench Science 0.1Artificial Analysis28.57%7of 24no list readGPT-6 Astra34.76 pts
ProgramBenchVals AI3.00%4of 294of 54Claude Opus 5.515.50 pts
τ³-BankingArtificial Analysis42.06%13of 30no list readQwen3.8 Max9.28 pts
AA-AnalystAgentArtificial Analysis53.75%3of 18no list readGemini 3.7 Flash6.25 pts
MLCR-AAArtificial Analysis55.56%5of 37no list readClaude Sonnet 5.519.44 pts
ARC-AGI-3ARC Prize · Standard30.16%2of 11no list readGPT-6 Astra32.55 pts
SWE-Bench Pro V2 FullScale AI / SEAL99.40%1of 10no list readClaude Opus 5—
SWE-Bench Pro V2 HARDScale AI / SEAL98.00%1of 10no list readClaude Opus 5—
Vals RSI IndexVals AI33.02%3of 203of 23Claude Opus 5.54.29 pts
Vibe Code Bench 1–100Vals AI28.53%2of 202of 20Claude Opus 5.51.83 pts
CUA-benchVals AI9.00%4of 84of 8GPT-6 Astra10.17 pts
Harvey LAB-AAArtificial Analysis93.46%5of 20no list readMuse Spark 1.3 Max2.02 pts
EnterpriseOps-Gym-AAArtificial Analysis47.48%6of 22no list readClaude Fable 53.64 pts
AutomationBench-AAArtificial Analysis56.57%24of 40no list readGemini 4 Argon20.94 pts
SciCodeArtificial Analysis56.37%19of 40no list readClaude Opus 5.510.53 pts
GDP.pdfArtificial Analysis21.60%18of 40no list readGPT-6.1 Sol9.40 pts
CritPtArtificial Analysis29.14%9of 42no list readGPT-5.6 Sol3.15 pts
AA-OmniscienceArtificial Analysis37.077of 42no list readClaude Opus 5.59.35
AA-LCR v1.1Artificial Analysis79.33%33of 42no list readKimi K39.34 pts
Terminal-Bench 2.1Vals AI84.64%5of 355of 75Claude Opus 5.53.00 pts
ProofBench v1.1Vals AI99.00%4of 345of 46Claude Sonnet 5.51.00 pts
LegalBenchVals AI86.97%7of 278of 149Claude Fable 51.59 pts
Harvey's Legal Agent BenchmarkVals AI6.67%18of 3724of 73Muse Spark 1.218.75 pts
SkillsBenchVals AI60.44%7of 208of 35DeepSeek V4.1 Flash9.36 pts
MysteryMechanismVals AI37.39%7of 217of 21GPT-6 Astra15.76 pts
Vals Multimodal IndexVals AI73.90%2of 172of 33Claude Fable 50.25 pts
BioMysteryBenchVals AI79.26%3of 193of 21Claude Sonnet 5.51.85 pts
Code MigrationVals AI57.47%6of 376of 71Claude Sonnet 5.512.36 pts
CyberBench v1.1Vals AI65.36%21of 3127of 43GPT-6 Sol12.62 pts
EMBVals AI73.56%6of 376of 69Claude Fable 5.13.11 pts
Finance Agent (v2)Vals AI58.63%7of 378of 73Gemini 4 Argon6.77 pts
IOIVals AI84.33%8of 318of 38Gemini 4 Argon15.67 pts
Legal Research BenchVals AI55.29%1of 371of 72Muse Spark 1.3 Max—
MedCodeVals AI63.57%1of 351of 104Claude Opus 5—
MedScribeVals AI90.98%4of 364of 106Claude Opus 5.50.45 pts
Public Benefits Bench v1.1Vals AI76.93%1of 291of 45Claude Opus 5—
SAGEVals AI49.43%11of 3221of 90Claude Opus 4.76.67 pts
SRE BenchVals AI12.21%8of 158of 16GPT-6 Astra44.66 pts
Tax Agent BenchVals AI75.06%3of 373of 64Claude Fable 5.12.58 pts
DeepSWE v1.1Datacurve73.65%3of 203of 28GPT-6 Astra0.47 pts
GDPval-AA v2.1Artificial Analysis1,707.924of 42no list readClaude Opus 5.5138.25
AA-Briefcase v1.1Artificial Analysis1,673.334of 41no list readClaude Opus 5.5148.52
Humanity's Last ExamArtificial Analysis54.87%6of 42no list readClaude Opus 5.56.48 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%Terminal-Bench Science 0.1%ProgramBench%
Claude Opus 5.5Anthropic66.97%5859.60%–––––90.29%59.05%18.50%
Claude Fable 5.1Anthropic65.83%5352.02%–90.52%93.43%92.38%90.64%90.26%43.33%7.00%
Claude Opus 5Anthropic63.67%5148.99%97.00%89.03%93.43%91.59%89.88%88.40%28.57%3.00%
GPT-6 AstraOpenAI63.13%5359.09%–––––89.59%63.33%5.50%
Claude Fable 5Anthropic61.39%5042.42%95.00%89.78%93.18%91.50%89.31%90.35%–2.00%
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
Claude Opus 5 (Adaptive Reasoning, Max Effort)$5.8656.4 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 analysis53.75%Artificial Analysisnot published
MLCR-AAMedical reasoning55.56%Artificial Analysisnot published
ARC-AGI-3Interactive reasoning30.16%ARC Prize · Standardnot published
SWE-Bench Pro V2 FullRepository coding99.40%Scale AI / SEALnot published
SWE-Bench Pro V2 HARDRepository coding98.00%Scale AI / SEALnot published
Vals RSI IndexAutonomous research33.02%Vals AInot published
Vibe Code Bench 1–100Application development28.53%Vals AInot published
CUA-benchComputer use9.00%Vals AInot published
Harvey LAB-AALegal reasoning93.46%Artificial Analysisnot published
EnterpriseOps-Gym-AAEnterprise workflows47.48%Artificial Analysisnot published
Apex Bench v1.1Reasoningno published result
65.8%Apex Bench v1.1DeepSeek · 17 October 2026Rank 2 of 6 in this publisher table
AutomationBench-AAWorkflow automation56.57%Artificial Analysisnot published
SciCodeScientific coding56.37%Artificial Analysisnot published
GDP.pdfDocument reasoning21.60%Artificial Analysisnot published
CritPtScientific reasoning29.14%Artificial Analysisnot published
AA-OmniscienceFactual knowledge37.07Artificial Analysisnot published
AA-LCR v1.1Long-context reasoning79.33%Artificial Analysisnot published
Terminal-Bench 4.0Agentic coding48.99%Artificial Analysis
53.53%Vals AI
51.8%Terminal-Bench 4.0DeepSeek · 17 October 2026Rank 1 of 6 in this publisher table
51.8%Terminal-Bench 4.0 (Pass@1)DeepSeek · 10 September 2026Rank 1 of 7 in this publisher table
52.6%Terminal-Bench 4.0OpenAI · 3 September 2026Rank 3 of 6 in this publisher table
52.3%Terminal-Bench 4.0Anthropic · 22 September 2026Rank 4 of 5 in this publisher table
Terminal-Bench 2.1Agentic coding84.64%Vals AI
89.14%Artificial Analysis
89.1%Terminal-Bench 2.1 (Pass@1)DeepSeek · 10 September 2026Rank 2 of 7 in this publisher table
SWE-bench (Verified)Software engineering97.00%Vals AInot published
LiveCodeBenchCode generation89.03%Vals AInot published
GPQA DiamondScience reasoning93.43%Vals AI
93.23%Artificial Analysis
93.4%GPQA Diamond (Pass@1)DeepSeek · 10 September 2026Rank 2 of 7 in this publisher table
93.7%GPQA DiamondOpenAI · 3 September 2026Rank 4 of 6 in this publisher table
MMLU ProKnowledge91.59%Vals AInot published
MMMU ProMultimodal reasoning89.88%Vals AI
84.74%Artificial Analysis
not published
Vibe Code Bench v1.1Product coding88.40%Vals AInot published
Terminal-Bench Science 0.1Agentic science28.57%Artificial Analysis
27.14%Vals AI
30.0%Terminal-Bench Science 0.1OpenAI · 3 September 2026Rank 3 of 5 in this publisher table
29.0%Terminal-Bench-Science 0.1Anthropic · 22 September 2026Rank 4 of 5 in this publisher table
ProgramBenchAgentic coding3.00%Vals AI
37.0%ProgramBenchDeepSeek · 17 October 2026Rank 1 of 6 in this publisher table
37.0%ProgramBench (Almost@1)DeepSeek · 10 September 2026Rank 1 of 6 in this publisher table
ProofBench v1.1Mathematics99.00%Vals AInot published
LegalBenchProfessional knowledge86.97%Vals AInot published
Harvey's Legal Agent BenchmarkProfessional knowledge6.67%Vals AInot published
SkillsBenchAgentic tooling60.44%Vals AInot published
MysteryMechanismScientific reasoning37.39%Vals AInot published
Vals Multimodal IndexComposite73.90%Vals AInot published
Vals IndexComposite63.67%Vals AInot published
BioMysteryBenchAgentic science79.26%Vals AInot published
Code MigrationAgentic coding57.47%Vals AInot published
CyberBench v1.1Security65.36%Vals AInot published
EMBFinancial modelling73.56%Vals AInot published
Finance Agent (v2)Financial analysis58.63%Vals AInot published
IOICompetitive programming84.33%Vals AInot published
Legal Research BenchProfessional knowledge55.29%Vals AInot published
MedCodeHealthcare administration63.57%Vals AInot published
MedScribeHealthcare administration90.98%Vals AInot published
Public Benefits Bench v1.1Public services76.93%Vals AInot published
SAGEEducation49.43%Vals AInot published
SRE BenchSecurity12.21%Vals AInot published
Tax Agent BenchProfessional knowledge75.06%Vals AInot published
OSWorld 2.0Computer useno published result
70.2%OSWorld 2.0 (offline set, partial reward)OpenAI · 3 September 2026Rank 2 of 3 in this publisher table
74.0%OSWorld 2.0 (partial)Anthropic · 22 September 2026Rank 3 of 3 in this publisher table
AutomationBenchBusiness workflowsno published result
50.3%AutomationBench (v1.0.6)DeepSeek · 17 October 2026Rank 3 of 6 in this publisher table
50.3%AutomationBench (Pass@1)DeepSeek · 10 September 2026Rank 2 of 7 in this publisher table
26.9%AutomationBench 1.0.6OpenAI · 3 September 2026Rank 3 of 5 in this publisher table
26.9%AutomationBenchAnthropic · 22 September 2026Rank 5 of 5 in this publisher table
CursorBench 4.0Agentic codingno published result
46.6%CursorBench 4.0Anthropic · 22 September 2026Rank 3 of 4 in this publisher table
DeepSWE v1.1Software engineering73.65%Datacurve
74.0%DeepSWE v1.1DeepSeek · 17 October 2026Rank 3 of 6 in this publisher table
74.0%DeepSWE v1.1 (Resolved)DeepSeek · 10 September 2026Rank 2 of 7 in this publisher table
73.7%DeepSWE v1.1OpenAI · 3 September 2026Rank 3 of 6 in this publisher table
FrontierCode v1.1Agentic codingno published result
53.4FrontierCode 1.1 MainOpenAI · 3 September 2026Rank 2 of 6 in this publisher table
48.0%FrontierCode v1.1 (Main)Anthropic · 22 September 2026Rank 4 of 5 in this publisher table
GDPval-AA v2.1Knowledge work1,707.92Artificial Analysis
1,708GDPval-AA v2.1Anthropic · 22 September 2026Rank 3 of 5 in this publisher table
AA-Briefcase v1.1Knowledge work1,673.33Artificial Analysisnot published
ChartographyVisual chart recognitionno published result
84.0%Chartography (w/tools)DeepSeek · 17 October 2026Rank 1 of 5 in this publisher table
84.0%Chartography w/ tools (Pass@1)DeepSeek · 10 September 2026Rank 1 of 4 in this publisher table
83.4%ChartographyAnthropic · 22 September 2026Rank 3 of 3 in this publisher table
Humanity's Last ExamMultidisciplinary reasoning54.87%Artificial Analysis
56.3%HLE (w/o tools)DeepSeek · 17 October 2026Rank 1 of 5 in this publisher table
56.3%HLE (Pass@1)DeepSeek · 10 September 2026Rank 1 of 7 in this publisher table
63.6%HLE w/ tools (Pass@1)DeepSeek · 10 September 2026Rank 2 of 6 in this publisher table
63.6%Humanity's Last Exam (with tools)OpenAI · 3 September 2026Rank 3 of 4 in this publisher table
63.6%Humanity's Last Exam (with tools)Anthropic · 22 September 2026Rank 3 of 4 in this publisher table
ARC-AGI-2Abstract reasoningno published result
90.4%ARC-AGI-2OpenAI · 3 September 2026Rank 3 of 5 in this publisher table
FrontierMath Tier 4 (v2)Mathematicsno published result
73.2%FrontierMath Tier 4 (v2)OpenAI · 3 September 2026Rank 5 of 5 in this publisher table
Agents' Last ExamComputer useno published result
28.6%Agents' Last ExamDeepSeek · 17 October 2026Rank 3 of 6 in this publisher table
28.6%Agents' Last Exam (Pass@1)DeepSeek · 10 September 2026Rank 2 of 7 in this publisher table
55.5%Agents' Last ExamOpenAI · 3 September 2026Rank 2 of 4 in this publisher table
Artificial Analysis Intelligence IndexComposite51Artificial Analysisnot published
Terminal-Bench 3.0Agentic codingno published result
43.3%Terminal-Bench 3.0 (Pass@1)DeepSeek · 10 September 2026Rank 1 of 7 in this publisher table
NL2Repo-BenchAgentic codingno published result
75.3%NL2Repo-BenchDeepSeek · 17 October 2026Rank 1 of 6 in this publisher table
75.3%NL2Repo-Bench (Score)DeepSeek · 10 September 2026Rank 1 of 7 in this publisher table
ExploitGymSecurityno published result
22.1%ExploitGym (Pass@1)DeepSeek · 10 September 2026Rank 2 of 6 in this publisher table
BabyVisionMultimodalno published result
94.1%BabyVision (w/tools)DeepSeek · 17 October 2026Rank 1 of 5 in this publisher table
94.1%BabyVision w/tools (Pass@1)DeepSeek · 10 September 2026Rank 1 of 4 in this publisher table
ZeroBench (main set)Multimodalno published result
52.0%ZeroBench-main (w/tools)DeepSeek · 17 October 2026Rank 2 of 5 in this publisher table
52.0%ZeroBench-main w/tools (Pass@5)DeepSeek · 10 September 2026Rank 2 of 4 in this publisher table
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.