Claude Opus 4

Anthropic · anthropic/claude-opus-4-20250514

ProviderAnthropic
Familyclaude-opus
Typellm-reasoning
Statusactive
Released2025-05-22
Updated2025-05-22

Reported benchmark scores

These scores carry one collection date for the whole card (2026-04) and a source list rather than a source per score, so they cannot be attributed individually. They are shown as reported and are not verified evidence. Benchmark pages carry the reviewed, dated evidence where it exists.
BenchmarkCatalogue standingScore
Aider Polyglot Benchmarkunassessed82.1
AIME 2025unassessed82.0
AlpacaEvalunassessed55.2
Arena Elo — Codingunassessed1420.0
Arena Elo — Hard Promptsunassessed1455.5
Arena Elo — Mathunassessed1400.0
Arena Elo — Overall (Text)unassessed1410.0
Arena Elo — Style Controlunassessed1423.7
Arena Elo — Visionunassessed1207.5
BBQ (Bias Benchmark for QA)unassessed88.2
ChartQAunassessed86.8
DocVQAunassessed93.2
FinBenchunassessed71.2
GPQA Diamondunassessed78.4
HELM Safetyunassessed92.5
HumanEvalunassessed93.2
IFEvalunassessed92.1
LegalBenchunassessed75.5
LiveCodeBenchunassessed62.4
MATH-500unassessed96.4
MathVistaunassessed65.5
MedQAunassessed82.1
MMLU: Astronomyunassessed80.5
MMLU: Biology (subcategory)unassessed88.2
MMLU: Business Ethicsunassessed82.1
MMLU: Chemistry (subcategory)unassessed82.5
MMLU: Clinical Knowledgeunassessed87.5
MMLU: Computer Science (subcategory)unassessed88.8
MMLU: Jurisprudenceunassessed80.5
MMLU: Physics (unresolved key)unassessed85.1
MMLU-Prounassessed85.2
MMLU: Professional Accountingunassessed72.5
MMLU: Professional Lawunassessed78.2
MMMUunassessed70.8
MT-Benchunassessed9.4
MultiPL-Eunassessed89.3
MultiPL-E: C++unassessed88.5
MultiPL-E: C#unassessed88.2
MultiPL-E: Gounassessed85.5
MultiPL-E: Javaunassessed91.2
MultiPL-E: JavaScriptunassessed91.5
MultiPL-E: Juliaunassessed70.2
MultiPL-E: Kotlin (unconfirmed)unassessed82.5
MultiPL-E: Luaunassessed65.8
MultiPL-E: Perlunassessed62.5
MultiPL-E: PHPunassessed85.5
MultiPL-E: Pythonunassessed95.5
MultiPL-E: Runassessed68.5
MultiPL-E: Rubyunassessed78.5
MultiPL-E: Rustunassessed82.1
MultiPL-E: Scalaunassessed72.2
MultiPL-E: Swiftunassessed76.8
MultiPL-E: TypeScriptunassessed90.8
SWE-bench Agentunassessed62.5
SWE-bench Verifiedunassessed72.7
τ-benchunassessed68.2
Terminal-Benchunassessed55.8
Terminal-Bench 2.0unassessed65.4
ToxiGenunassessed95.1
WildBenchunassessed82.5

Data

This card as JSON · Edit on GitHub