Prior Opus-tier flagship preceding Opus 4.7, now listed as a legacy model. Offers a 1M-token context window, extended and adaptive thinking, and full vision/tool-use support at the standard Opus price of $5/$25 per million tokens.
profile normalized against the 47-model field
Context window· 1M of 1.05M95%
Max output· 128K of 272K47%
Output speed· 62 tok/s19%
Affordability· $25.00 / Mtok out86%
Capability breadth· 4 of 1136%
Capability switches · 4 of 11
Reasoning mode
Tool / function use
Streaming
JSON mode
Structured outputs
Prompt caching
Fine-tuning
Web search
Code execution
Vision input
Audio input
Specifications
Every value carries a primary source and a verification date.
Capacity
Context window
1M
Max output
128K
Pricing
Input $/Mtok
$5.00 / Mtok input USD/Mtok
Cached input $/Mtok
0.5 USD/Mtok
Output $/Mtok
$25.00 / Mtok output USD/Mtok
Performance
Output speed
62 tok/s tok/s
Capabilities
Input modalities
Text, image
Output modalities
Text
Reasoning mode
Extended + adaptive thinking: Yes
Tool / function use
Yes
Prompt caching
Yes
Vision input
Yes
API
API model ID
claude-opus-4-6
Batch API
Yes
General
Knowledge cutoff
May 2025
Release date
March 12, 2026
Benchmarks
Sourced evaluation scores, each verified against its primary source.
MMLU-Pro
MMLU-Pro: 87.9% for Claude Opus 4.6.
87.9 %Verified
GPQA Diamond
Claude Opus 4.6 reaches GPQA Diamond: 85.4% on graduate-level science questions.
85.4 %Verified
SWE-bench Verified
On SWE-bench Verified, Claude Opus 4.6 scores 79.8%.
79.8 %Verified
Humanity's Last Exam (no tools)
Humanity's Last Exam (no tools): 18.6% for Claude Opus 4.6.
18.6 %Verified
AIME 2025
AIME 2025: 90.5% with extended thinking enabled.
90.5 %Verified
Terminal-Bench 2.1
Terminal-Bench 2.1: 51.2% on agentic terminal tasks.