AIEvals

Models

Claude Opus 4.8

Specifications and API prices for Claude Opus 4.8, recorded by Artificial Analysis.

AnthropicReleased 28 May 2026Context 1M tokensOutput Not published in the checked sourcePrivate weights$5.00 in / $25.00 out per Mtok

Family Claude Opus55 benchmarks with a published resultNo first place on the independent columns 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 AI55.10%13of 3313of 41Gemini 4 Argon13.80 pts
Artificial Analysis Intelligence IndexArtificial Analysis4220of 42no list readClaude Opus 5.516 pts
Terminal-Bench 4.0Artificial Analysis21.72%23of 40no list readClaude Sonnet 5.541.92 pts
SWE-bench (Verified)Vals AI88.60%11of 2312of 88Claude Opus 58.40 pts
LiveCodeBenchVals AI87.82%9of 2111of 143Claude Fable 5.12.70 pts
GPQA DiamondVals AI92.42%14of 2316of 138Gemini 3.1 Pro3.03 pts
MMLU ProVals AI89.59%8of 249of 138Claude Fable 5.12.79 pts
MMMU ProVals AI86.59%11of 2018of 93Claude Fable 5.14.05 pts
Vibe Code Bench v1.1Vals AI82.72%15of 3816of 106Claude Sonnet 5.59.67 pts
ProgramBenchVals AI1.00%11of 2911of 54Claude Opus 5.517.50 pts
τ³-BankingArtificial Analysis34.23%24of 30no list readQwen3.8 Max17.11 pts
IFBenchArtificial Analysis62.24%9of 10no list readGrok 4 20 030920.69 pts
AA-AnalystAgentArtificial Analysis45.00%9of 18no list readGemini 3.7 Flash15.00 pts
MLCR-AAArtificial Analysis45.56%9of 37no list readClaude Sonnet 5.529.44 pts
Vals RSI IndexVals AI23.98%9of 209of 23Claude Opus 5.513.33 pts
τ²-Bench TelecomArtificial Analysis94.44%3of 10no list readJt 35b Flash4.68 pts
Harvey LAB-AAArtificial Analysis91.08%10of 20no list readMuse Spark 1.3 Max4.40 pts
EnterpriseOps-Gym-AAArtificial Analysis43.96%12of 22no list readClaude Fable 57.16 pts
AutomationBench-AAArtificial Analysis45.59%33of 40no list readGemini 4 Argon31.92 pts
SciCodeArtificial Analysis54.40%25of 40no list readClaude Opus 5.512.50 pts
GDP.pdfArtificial Analysis22.80%13of 40no list readGPT-6.1 Sol8.20 pts
CritPtArtificial Analysis20.86%17of 42no list readGPT-5.6 Sol11.43 pts
AA-OmniscienceArtificial Analysis28.7513of 42no list readClaude Opus 5.517.67
AA-LCR v1.1Artificial Analysis77.67%38of 42no list readKimi K311.00 pts
Terminal-Bench 2.1Vals AI71.91%22of 3525of 75Claude Opus 5.515.73 pts
LegalBenchVals AI83.57%24of 2751of 149Claude Fable 54.99 pts
Harvey's Legal Agent BenchmarkVals AI9.58%13of 3716of 73Muse Spark 1.215.84 pts
SkillsBenchVals AI59.23%8of 209of 35DeepSeek V4.1 Flash10.57 pts
Vals Multimodal IndexVals AI70.89%5of 175of 33Claude Fable 53.26 pts
Code MigrationVals AI47.25%14of 3714of 71Claude Sonnet 5.522.58 pts
EMBVals AI69.37%13of 3713of 69Claude Fable 5.17.30 pts
Finance Agent (v2)Vals AI53.92%19of 3724of 73Gemini 4 Argon11.48 pts
Legal Research BenchVals AI43.75%17of 3718of 72Muse Spark 1.3 Max11.54 pts
MedCodeVals AI53.22%8of 3510of 104Claude Opus 510.35 pts
MedScribeVals AI85.75%19of 3624of 106Claude Opus 5.55.68 pts
Public Benefits Bench v1.1Vals AI68.13%11of 2911of 45Claude Opus 58.80 pts
SAGEVals AI54.79%2of 323of 90Claude Opus 4.71.31 pts
Tax Agent BenchVals AI64.91%17of 3719of 64Claude Fable 5.112.73 pts
DeepSWE v1.1Datacurve58.97%15of 2015of 28GPT-6 Astra15.15 pts
GDPval-AA v2.1Artificial Analysis1,438.3231of 42no list readClaude Opus 5.5407.85
AA-Briefcase v1.1Artificial Analysis1,320.831of 41no list readClaude Opus 5.5501.05
Humanity's Last ExamArtificial Analysis48.66%12of 42no list readClaude Opus 5.512.69 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%ProgramBench%
GPT-6 SolOpenAI57.54%4843.94%–––––87.82%2.00%
MiMo V2.6 ProXiaomi55.20%4634.85%–––––85.22%0.50%
Claude Opus 4.8Anthropic55.10%4221.72%88.60%87.82%92.42%89.59%86.59%82.72%1.00%
Grok 4.7xAI54.95%4625.76%–––––86.17%0.50%
Gemini 3.8 FlashGoogle54.83%4119.70%80.00%89.48%94.44%90.22%89.08%78.65%1.00%
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 4.8 (Adaptive Reasoning, Max Effort)$4.0856.3 tok/s

Results by benchmark

Independent and vendor results retain their own harnesses and effort settings.

BenchmarkIndependent runAs reported by a publisher
τ³-BankingTool use34.23%Artificial Analysisnot published
IFBenchInstruction following62.24%Artificial Analysisnot published
AA-AnalystAgentData analysis45.00%Artificial Analysisnot published
MLCR-AAMedical reasoning45.56%Artificial Analysisnot published
Vals RSI IndexAutonomous research23.98%Vals AInot published
τ²-Bench TelecomTool use94.44%Artificial Analysisnot published
Harvey LAB-AALegal reasoning91.08%Artificial Analysisnot published
EnterpriseOps-Gym-AAEnterprise workflows43.96%Artificial Analysisnot published
AutomationBench-AAWorkflow automation45.59%Artificial Analysisnot published
SciCodeScientific coding54.40%Artificial Analysisnot published
GDP.pdfDocument reasoning22.80%Artificial Analysisnot published
CritPtScientific reasoning20.86%Artificial Analysisnot published
AA-OmniscienceFactual knowledge28.75Artificial Analysisnot published
AA-LCR v1.1Long-context reasoning77.67%Artificial Analysisnot published
Terminal-Bench 4.0Agentic coding21.72%Artificial Analysis
23.23%Vals AI
not published
Terminal-Bench 2.1Agentic coding71.91%Vals AI
84.64%Artificial Analysis
85.0%Terminal Bench 2.1Z.ai · 14 August 2026Rank 7 of 8 in this publisher table
SWE-bench (Verified)Software engineering88.60%Vals AInot published
LiveCodeBenchCode generation87.82%Vals AInot published
GPQA DiamondScience reasoning92.42%Vals AI
92.02%Artificial Analysis
not published
MMLU ProKnowledge89.59%Vals AInot published
MMMU ProMultimodal reasoning86.59%Vals AInot published
Vibe Code Bench v1.1Product coding82.72%Vals AInot published
Terminal-Bench Science 0.1Agentic science
4.29%Vals AI
not published
ProgramBenchAgentic coding1.00%Vals AI
15.5%ProgramBench (Almost Solved)Z.ai · 14 August 2026Rank 5 of 7 in this publisher table
LegalBenchProfessional knowledge83.57%Vals AInot published
Harvey's Legal Agent BenchmarkProfessional knowledge9.58%Vals AInot published
SkillsBenchAgentic tooling59.23%Vals AInot published
Vals Multimodal IndexComposite70.89%Vals AInot published
Vals IndexComposite55.10%Vals AInot published
Code MigrationAgentic coding47.25%Vals AInot published
EMBFinancial modelling69.37%Vals AInot published
Finance Agent (v2)Financial analysis53.92%Vals AInot published
Legal Research BenchProfessional knowledge43.75%Vals AInot published
MedCodeHealthcare administration53.22%Vals AInot published
MedScribeHealthcare administration85.75%Vals AInot published
Public Benefits Bench v1.1Public services68.13%Vals AInot published
SAGEEducation54.79%Vals AInot published
Tax Agent BenchProfessional knowledge64.91%Vals AInot published
AutomationBenchBusiness workflowsno published result
41.0%AutomationBench (v1.0.6)Z.ai · 14 August 2026Rank 6 of 8 in this publisher table
DeepSWE v1.1Software engineering58.97%Datacurve
58.0%DeepSWE (v1.1)Z.ai · 14 August 2026Rank 6 of 8 in this publisher table
GDPval-AA v2.1Knowledge work1,438.32Artificial Analysisnot published
AA-Briefcase v1.1Knowledge work1,320.8Artificial Analysisnot published
Humanity's Last ExamMultidisciplinary reasoning48.66%Artificial Analysis
57.9%HLE w/ ToolsZ.ai · 14 August 2026Rank 6 of 8 in this publisher table
Agents' Last ExamComputer useno published result
25.7%Agents' Last Exam (ALE-CLI)Z.ai · 14 August 2026Rank 5 of 8 in this publisher table
Artificial Analysis Intelligence IndexComposite42Artificial Analysisnot published
Terminal-Bench 3.0Agentic codingno published result
21.1%Terminal Bench 3.0Z.ai · 14 August 2026Rank 4 of 6 in this publisher table
NL2Repo-BenchAgentic codingno published result
69.7%NL2RepoZ.ai · 14 August 2026Rank 1 of 6 in this publisher table
CyberGymSecurityno published result
78.1%CyberGymZ.ai · 14 August 2026Rank 7 of 8 in this publisher table
ExploitGymSecurityno published result
80ExploitGym (2h)Z.ai · 14 August 2026Rank 4 of 7 in this publisher table
120ExploitGym (6h)Z.ai · 14 August 2026Rank 4 of 7 in this publisher table
FrontierSWEAgentic researchno published result
66.5%FrontierSWEZ.ai · 14 August 2026Rank 4 of 4 in this publisher table
SWE-MarathonAgentic codingno published result
48.8%SWE-Marathon (v1.1)Z.ai · 14 August 2026Rank 1 of 6 in this publisher table
PostTrainBenchAI researchno published result
32.9%PostTrainBenchZ.ai · 14 August 2026Rank 4 of 6 in this publisher table
ExploitBenchSecurityno published result
40.0%ExploitBenchZ.ai · 14 August 2026Rank 4 of 7 in this publisher table
Toolathlon VerifiedAgentic toolingno published result
76.2%Toolathlon VerifiedZ.ai · 14 August 2026Rank 2 of 8 in this publisher table
GDPval-AA v2Knowledge workno published result
1,588GDPval-AA v2Z.ai · 14 August 2026Rank 7 of 8 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.