Model
Model explorer

Claude Opus 4.5 (High)

CLOSED
Anthropic · Claude 4 family · released Nov 24, 2025

Efficiency-focused flagship; API default (high) effort; record 80.9% SWE-bench Verified with major price cut vs Opus 4.1.

ReasoningCodingVisionFunction callingTool useAgentic
2301.0
Elo · rank #34
Parameters
Undisclosed
Active params
Context
200K tokens
Architecture
Dense Transformer (undisclosed configuration)
License
Proprietary
Languages
API price (in/out)
$5 / $25
Modalities
text · vision
Benchmark results
Bar shows position within the tracked field; marker = field best
AIMEMath92.8%#32
best: GPT-5.2 · 100.0%
ARC-AGI-1Reasoning80.0%#7
best: GPT-5.6 · 96.5%
ARC-AGI-2Reasoning37.6%#14
best: GPT-5.6 · 92.5%
CyberGymCoding50.6%#7
best: GPT-5.6 · 84.5%
GPQA DiamondReasoning87.0%#45
best: GPT-5.6 · 94.6%
Humanity's Last ExamReasoning43.2%#17
best: Claude Sonnet 5 · 57.4%
LiveCodeBenchCoding83.7%#31
best: DeepSeek-V4-Pro (Think Max) · 93.5%
MMLU-ProKnowledge87.3%#11
best: Claude Fable 5 · 91.5%
MMMLUKnowledge90.8%#5
best: Gemini 3.1 Pro · 92.6%
MMMUVision80.7%#23
best: Claude Fable 5 · 89.3%
OSWorld-VerifiedAgents66.3%#16
best: Claude Fable 5 · 85.0%
SWE-bench ProCoding52.0%#36
best: Claude Fable 5 · 80.0%
SWE-bench VerifiedCoding80.9%#10
best: Claude Fable 5 · 95.0%
τ²-Bench RetailAgents88.9%#3
best: Claude Sonnet 4.6 · 91.7%
τ²-Bench TelecomAgents98.2%#6
best: Claude Opus 4.6 · 99.3%
Terminal-Bench 2.0Coding59.3%#33
best: GPT-5.6 · 88.8%
Terminal-BenchCoding59.3%#1
best: this model · 59.3%
API price $5/$25 · each benchmark row carries its own source badge (see methodology)