Inference
EN
Anthropic

Claude Opus 4.5

Claude Opus 4.5 is a standard model from Anthropic. 60 benchmarks count toward its index, in 8 categories.

  • Closed weights
  • Standard
  • Input: file · image · text
  • Released 2025-11
RankedEnough results for an index.
IndexAll categories, out of 100.52.0±2.1Median 51.0
CoverageIndex weight that has results.100%
SpeedOutput tokens per second.39/sMedian 67/s
First tokenSeconds to the first output token.1.1 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$5.00batch $2.50Median $0.60
Output / 1MUS dollars per 1M output tokens.$25.00batch $12.50 Batch price per 1M output tokens.Median $2.55
ContextMaximum tokens in one request.200KMedian 262K
EloLMArena rating and rank.1451±3 #52Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
anthropic/claude-opus-4.5
Max outputMaximum output tokens per request.
64K
Knowledge cutoffLast date of training data.
—
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.50
WeightsLink to the published weights.
—
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
6 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
AnthropicMaker$5.00$25.00200K
DatabricksCloud$5.00$25.00200K
DigitalOceanCloud$5.00$25.00200K
OpenRouterRouter$5.00$25.00200K
SAP AI CoreCloud$5.00$25.00200K
Snowflake CortexCloud$5.00$25.00200K

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.15 counted54.7105 / 218
CodingCode writing and repair.6 counted55.098 / 209
ReasoningLogic problems and puzzles.7 counted51.8103 / 227
MultimodalTasks with images and text.10 counted42.273 / 84
KnowledgeFacts and expert knowledge.8 counted54.698 / 268
MultilingualTasks in many languages.2 counted56.62 / 9
InstructionTasks with strict prompt rules.2 counted41.243 / 63
MathMath problems.10 counted51.380 / 168

Results

60 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
SWE-bench Verified · SWE-bench/experimentsCoding76.8%1 / 65
Massive Multitask Language Understanding ProfessionalKnowledge89.5%2 / 75
τ²-bench AirlineAgentic84.0%1 / 22
MMLU-ReduxKnowledge96.6%1 / 19
Software Engineering Benchmark VerifiedCoding80.9%13 / 114
Gert Labs Composite Game BenchmarkAgentic64.2%6 / 52
MMLU-ProXMultilingual85.7%2 / 17
Terminal-Bench 1.0Agentic56.3% ±5.57 / 47
Terminal-Bench 2.0 · Vals AIAgentic58.4% ±5.310 / 66
τ³-Bench Tool-Agent-User EvaluationAgentic70.2%7 / 32
LongBench v2Reasoning64.4%6 / 25
INCLUDEMultilingual86.2%3 / 12
SWE-Bench verified · Epoch AICoding76.7% ±1.99 / 32
Artificial Analysis Omniscience AccuracyKnowledge40.9%47 / 164
IOI v1Coding23.6% ±7.519 / 61
Graduate-Level Google-Proof Q&AKnowledge87.0%44 / 141
GPQA diamond · Epoch AIKnowledge85.5% ±2.270 / 217
τ²-bench TelecomAgentic92.3%6 / 18
C-EvalKnowledge92.2%4 / 12
τ²-Bench Tool-Agent-User EvaluationAgentic86.3%51 / 149
Instruction-Following EvalInstruction90.9%24 / 70
SuperGPQA: Scaling LLM Evaluation Across 285 Graduate DisciplinesKnowledge70.6%8 / 22
LiveCodeBench v6Coding84.8%22 / 59
Massive Multi-discipline Multimodal UnderstandingMultimodal80.7%11 / 28
MMLU ProKnowledge85.6% ±0.455 / 135
OTIS Mock AIME 2024-2025Math81.7% ±5.083 / 200
Harvard-MIT Mathematics Tournament November 2025Math93.3%5 / 12
FrontierMath-2025-02-28-PrivateMath20.7% ±2.430 / 69
Terminal-Bench 2.0 · BenchLMAgentic59.3%18 / 41
MMMU ProMultimodal81.1% ±0.942 / 92
SimpleQA VerifiedKnowledge45.7% ±1.642 / 91
AIME 2026Math95.1%16 / 34
FrontierMath-Tier-4-2025-07-01-PrivateMath4.2% ±2.926 / 55
τ²-bench RetailAgentic79.6%11 / 23
Artificial Analysis GPQA DiamondKnowledge81.0%96 / 199
WideResearchAgentic76.4%8 / 16
Artificial Analysis Long Context ReasoningReasoning70.7%110 / 217
GPQA Diamond · Vals AIKnowledge79.5% ±2.569 / 135
Mystery Game PuzzlesReasoning22.0% ±4.241 / 80
AIMEMath76.9% ±1.450 / 95
SWE-bench ProCoding57.1%45 / 83
LiveCodeBenchCoding75.0% ±1.180 / 142
NL2RepoCoding43.2%21 / 36
Humanity's Last ExamKnowledge30.8%45 / 77
Harvard-MIT Mathematics Tournament February 2026Math85.3%16 / 27
Artificial Analysis Humanity's Last ExamKnowledge13.2%132 / 222
Harvard-MIT Mathematics Tournament February 2025Math92.9%9 / 15
SimpleVQAMultimodal65.7%11 / 18
Artificial Analysis MMMU-ProMultimodal71.2%72 / 113
Critical Physics TasksReasoning0.3%137 / 214
Artificial Analysis IFBenchInstruction43.0%94 / 146
Claw-EvalAgentic59.6%28 / 42
OSWorld-VerifiedAgentic66.3%25 / 37
ToolathlonAgentic43.5%15 / 22
VITA-BenchAgentic23.3%9 / 13
JobBenchAgentic32.3%25 / 36
CountBenchMultimodal90.6%7 / 10
FrontierMath-Tiers-1-3-v2-PrivateMath34.4% ±2.863 / 90
EBR-benchReasoning14.3%22 / 31
τ²-bench BankingAgentic24.7%21 / 28
Furniture AssemblyReasoning28.3% ±5.928 / 37
RealWorldQAMultimodal77.0%14 / 18
Instruction Following BenchmarkInstruction58.0%45 / 57
Massive Multi-discipline Multimodal Understanding ProMultimodal70.6%44 / 53
FrontierMath-Tier-4-v2-PrivateMath4.9% ±3.462 / 72
QwenClawBenchAgentic52.3%9 / 10
CyberGymAgentic50.6%27 / 29
MCP AtlasAgentic42.3%49 / 52
Chess PuzzlesReasoning5.0% ±2.285 / 90
MathVisionMultimodal74.3%23 / 24
CharXiv ReasoningMultimodal68.5%47 / 49
V*Multimodal67.0%14 / 14
ERQAMultimodal46.8%18 / 18
ScreenSpot ProMultimodal45.7%20 / 20
MGSMMultilingual94.8% ±1.42 / 74
MortgageTaxFinance68.7% ±0.911 / 97
SWE-bench (full test split)Coding52.6%1 / 7
SWE-bench Multilingual · SWE-bench/experimentsCoding70.7%3 / 14
MedQAHealthcare93.2% ±0.224 / 95
TaxEval v2Finance74.3% ±0.938 / 143
FrontierMath-2025-02-28-PublicMath30.0% ±16.322 / 64
MedCodeHealthcare45.2% ±1.938 / 106
CorpFin v2Finance61.3% ±1.054 / 131
MedScribeHealthcare83.2% ±1.947 / 109
LegalBenchLegal82.8% ±0.465 / 149
OmniDocBench 1.5Multimodal87.7%—
AI2D test splitMultimodal87.7%—
VideoMMMUMultimodal84.4%—
MMAnswerBenchMath84.0%—
MLVU mean averageMultimodal81.7%—
Video-MME without subtitleMultimodal81.4%—
DynaMathMultimodal79.7%—
MAXIFEMultilingual79.2%—
PolyMathMultilingual79.0%—
Video-MME with subtitleMultimodal77.6%—
AI-NeedleReasoning74.0%—
MCP-TasksAgentic71.8%—
We-MathMultimodal70.0%—
OSWorldAgentic66.3%—
NOVA-63Multilingual56.7%—
DeepPlanningAgentic26.4%—

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseVals AINo licence statedSierra ResearchMITEpoch AICC BYSWE-bench teamNo licence statedmodels.devMIT

History

5 releases
Index52.9 → 52.0
23 Oct 202612 Nov 2026
Output / 1M$25.00 → $25.00
23 Oct 202612 Nov 2026

Questions

How much does Claude Opus 4.5 cost?
$5.00 per 1M input tokens, $25.00 per 1M output tokens.
How good is Claude Opus 4.5?
Its index is 52.0 out of 100. That is place 131 of 283 ranked models.
What is Claude Opus 4.5 best at?
Multilingual: 56.6 out of 100, place 2 of 9.
How fast is Claude Opus 4.5?
39 output tokens per second, 1.1 s to the first token.
What is the context window of Claude Opus 4.5?
200K tokens.
Is Claude Opus 4.5 open source?
No. Its weights are not published.
When was Claude Opus 4.5 released?
November 2025.

Badge

Claude Opus 4.5 on Inference360: place 131 of 283, index 52.0
HTML
<a href="https://inference360.ai/models/claude-opus-4-5"><img src="https://inference360.ai/badges/claude-opus-4-5.svg" alt="Claude Opus 4.5 on Inference360: place 131 of 283, index 52.0" height="20"></a>
Markdown
[![Claude Opus 4.5 on Inference360: place 131 of 283, index 52.0](https://inference360.ai/badges/claude-opus-4-5.svg)](https://inference360.ai/models/claude-opus-4-5)

Compare with

Gemini 3.5 Flash-Lite52.0Qwen3.6-27B52.1A.X K251.9GLM-5-Turbo51.9Claude Opus 4.5 Thinking56.3

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from Anthropic

All models
Claude Opus 4.5 Thinking56.3Claude Haiku 4.5 Thinking43.5Claude Haiku 4.543.2Claude Sonnet 4.5 Thinking53.8Claude Sonnet 4.547.9