Anthropic

Claude Opus 4.5

Multimodal
#1SWE Bench Verified
#1MCP-Atlas
#2GDPVal
+1 more

by Anthropic

+
+
+
+
About

Claude Opus 4.5, released by Anthropic in November 2025, is optimized for long-horizon autonomous tasks — scenarios requiring the model to maintain a clear goal, reason carefully through many sequential steps, and produce reliable outputs over extended sessions without human guidance. Anthropic specifically designed it for multi-agent orchestration, where it serves as an orchestrator capable of decomposing complex tasks, delegating to specialized sub-agents, and synthesizing results across long sessions.

+
+
+
+
Pricing Range
Input (per 1M)$5.00 -$5.00
Output (per 1M)$25.00 -$25.00
Providers3
+
+
+
+
Timeline
ReleasedNov 24, 2025
Knowledge CutoffMay 1, 2025
+
+
+
+
Specifications
Capabilities
Multimodal
+
+
+
+
License & Family
License
Proprietary
Performance Overview
Performance metrics and category breakdown

Overall Performance

9 benchmarks
Average Score
46.5%
Best Score
80.9%
High Performers (80%+)
1

Performance Metrics

Max Context Window
264.0K

Top Categories

Tool Use
62.3%
Coding
61.5%
Finance
58.8%
Agents
45.5%
Reasoning
34.2%
+
+
+
+
All Benchmark Results for Claude Opus 4.5
Complete list of benchmark scores with detailed information
SWE Bench Verified
Coding
80.90
80.9%
Unverified
MCP-Atlas
Tool Use
62.30
62.3%
Unverified
Terminal Bench 2.0
Coding
59.80
59.8%
Unverified
Finance Agent
Finance
58.80
58.8%
Unverified
GDPVal
Agents
45.50
45.5%
Unverified
SWE-rebench
Coding
43.80
43.8%
Unverified
ARC-AGI-2
Reasoning
37.60
37.6%
Unverified
Humanity's Last Exam
Reasoning
30.80
30.8%
Unverified
MMMU
Multimodal
-1.00
-1.0%
Unverified