Claude Sonnet 4.6 is Anthropic's latest Sonnet-class model released in February 2026, delivering frontier performance across coding, agents, and professional work at the Sonnet price point. It features a 1M-token context window with extended thinking and 64K max output tokens, and introduces the effort parameter for fine-tuning the balance between speed, cost, and reasoning depth. The model excels at iterative development, complex codebase navigation, polished document creation, and confident computer use for web QA and workflow automation.
Anthropic FreeAnthropic ProAnthropic Max (5x)Anthropic Max (20x)API|VisionReasoningWeb Search|Proprietary Model
Knowledge Cutoff
2025-08-01
Input → Output Format
Context Memory
1MIN64KOUT
AI Performance Evaluation
Arena Overall Score
1463
±6As of 2026-04-23
Overall Rank
No.21
13,434 Votes
Arena by Ability
Hard Prompts
1498±7No.11
Expert Knowledge
1502±18No.11
Instruction Following
1475±10No.9
Conversation Memory
1483±12No.15
Creative
1444±13No.23
Coding
1520±11No.11
Math
1465±19No.22
Arena by Occupation
Creative Writing
1453±11No.16
Social Sciences
1494±13No.8
Media
1443±12No.17
Business
1480±12No.9
Healthcare
1469±19No.38
Legal
1474±19No.24
Software
1509±9No.14
Mathematics
1489±22No.8
Source:Arena Intelligence
Overall
AA Intelligence Index
43%↑4%
LiveBench
70%↑10%
ForecastBench
59%↑0%
Reasoning & Math
GPQA Diamond
80%↓1%
HLE
11%↓6%
LB Reasoning
77%↑18%
LB Math
80%↑7%
LB Data
75%↑25%
Coding
AA Coding Index
43%↑9%
LB Coding
74%↑1%
LB Agentic
63%↑20%
TAU2
79%↑6%
TerminalBench
42%↑11%
SciCode
44%↑3%
Language & Instructions
IFBench
42%↓14%
AA-LCR
59%↓3%
Hallucination (HHEM)
11%↑0%
Factual (HHEM)
89%↑0%
LB Language
71%↑0%
LB IF
52%↑6%
Output Speed
Standard Mode
59tok/s↓23
First Output 1.08s
Reasoning Mode
50tok/s↓38
First Output 32.71s
