evals.report
BenchmarksLabsCompareRun guidesIn the wild
AnthropicClaude Opus

Claude Opus 4.5

Anthropic · Claude Opus. Released Nov 24, 2025.

Claude Opus 4.5 is a model from Anthropic in the Claude Opus family, released Nov 24, 2025. evals.report tracks 46 reported Claude Opus 4.5 benchmark scores across FrontierMath, Berkeley Function Calling Leaderboard, ARC-AGI-1, LiveBench, SWE-bench Pro, GPQA Diamond, SWE-bench Verified, Humanity's Last Exam, and 38 more — each shown with its benchmark, metric, source status, and date, and never combined into a single ranking.

46 results

Benchmark results 46

Compare this model
BenchmarkCategoryScoreMetricStatusDate
FrontierMathReasoning20.69%accuracyOfficialNov 24, 2025Details
Berkeley Function Calling LeaderboardTool use77.47%accuracyOfficialNov 24, 2025Details
ARC-AGI-1Reasoning80%accuracyOfficialNov 24, 2025Details
LiveBenchReasoning75.96%scoreOfficialNov 24, 2025Details
SWE-bench ProCoding45.89%% resolvedOfficialNov 24, 2025Details
GPQA DiamondReasoning86.0%accuracyOfficialNov 24, 2025Details
SWE-bench VerifiedCoding76.7%% resolvedOfficialNov 24, 2025Details
Humanity's Last ExamReasoning25.8%accuracyOfficialNov 24, 2025Details
MMMU-ProMultimodal73.9%accuracyOfficialNov 24, 2025Details
AIME (OTIS Mock)Reasoning86.1%accuracyOfficialNov 24, 2025Details
SimpleQA VerifiedOther41.8%accuracyOfficialNov 24, 2025Details
ARC-AGI-2Reasoning37.64%accuracyOfficialNov 24, 2025Details
LMArenaChat preference1449source-defined ratingOfficialNov 24, 2025Details
WeirdMLCoding63.7%average accuracyOfficialNov 24, 2025Details
MCP AtlasTool use62.3%pass rateOfficialNov 24, 2025Details
PostTrainBenchAgents17.29%weighted average scoreOfficialNov 24, 2025Details
Artificial Analysis Intelligence IndexReasoning49.7IndexUnverifiedNov 24, 2025Details
Epoch Capabilities IndexReasoning149.9IndexOfficialNov 24, 2025Details
Aider PolyglotCoding89.4%% correctVerifiedNov 24, 2025Details
MMLU-ProReasoning89.5%accuracyVerifiedNov 24, 2025Details
OSWorldAgents66.3%task success rateUnverifiedNov 24, 2025Details
τ²-bench (Telecom)Tool use98.2%pass^1UnverifiedNov 24, 2025Details
GDPvalAgents1452EloOfficialNov 24, 2025Details
LiveCodeBenchCoding73.8%Pass@1UnverifiedNov 24, 2025Details
METR Task-Completion Time HorizonsAgents293.0 min50% time horizonOfficialNov 24, 2025Details
SciCodeCoding49.5%accuracyUnverifiedNov 24, 2025Details
MMMU (Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark)Multimodal80.7%accuracyVerifiedNov 24, 2025Details
MultiChallengeReasoning58.97%accuracyVerifiedNov 24, 2025Details
Global-MMLUReasoning91.3%accuracyUnverifiedNov 24, 2025Details
WebDev ArenaChat preference1467EloVerifiedNov 24, 2025Details
Search ArenaChat preference1182EloVerifiedNov 24, 2025Details
EQ-Bench Creative Writing v3Chat preference1762EloVerifiedNov 24, 2025Details
Design ArenaChat preference1295EloVerifiedNov 24, 2025Details
MASK (Model Alignment between Statements and Knowledge)Other92.53Honesty scoreVerifiedNov 24, 2025Details
ScreenSpot-ProMultimodal45.7%accuracyUnverifiedNov 24, 2025Details
FACTS GroundingReasoning62.1%Grounding accuracyVerifiedNov 24, 2025Details
SWE-bench MultilingualCoding70.7%% resolvedOfficialNov 24, 2025Details
EnigmaEvalReasoning11.91%accuracyVerifiedNov 24, 2025Details
ZeroBenchMultimodal10.0% (pass@5)accuracyVerifiedNov 24, 2025Details
FrontierMath Tier 4Reasoning4.2%accuracyOfficialNov 24, 2025Details
Vectara Hallucination LeaderboardOther10.9%Hallucination RateOfficialNov 24, 2025Details
Gray Swan Arena (Agent Red-Teaming / Indirect Prompt Injection)Agents0.5%Attack Success Rate (ASR)VerifiedNov 24, 2025Details
Vibe Code BenchCoding20.63%Overall accuracyVerifiedNov 24, 2025Details
WebArenaAgents65.3%Task success rateVerifiedNov 24, 2025Details
GSO: Software Optimization Benchmark for SWE-AgentsCoding26.47%Opt@1OfficialNov 24, 2025Details
MultiNRCReasoning48.63%accuracyOfficialNov 24, 2025Details