TAU2
Measured May 14, 2026Source
Score
0.76
Models9 1 
Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)
AnthropicAnthropicUnited States
Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort) is a balanced model in the Claude family, optimized for a strong mix of performance, speed, and cost-efficiency. The 'Adaptive Reasoning' and 'Max Effort' designations suggest enhanced capabilities for complex reasoning tasks, potentially with dynamic resource allocation. It is well-suited for applications requiring reliable intelligence without the premium cost of the largest models.
ReasoningFastCheapLong context
Input / 1M tokens
$3.75Output / 1M tokens
$15.00Output tokens/s
68.41First-token seconds
58.05sSupported plans
6Benchmark history
Evaluations
Terminalbench Hard
Measured May 14, 2026Source
Score
0.53
Lcr
Measured May 14, 2026Source
Score
0.71
Ifbench
Measured May 14, 2026Source
Score
0.57
Scicode
Measured May 14, 2026Source
Score
0.47
Hle
Measured May 14, 2026Source
Score
0.3
Gpqa
Measured May 14, 2026Source
Score
0.88
Artificial Analysis Coding Index
Measured May 14, 2026Source
Score
50.9
Artificial Analysis Intelligence Index
Measured May 14, 2026Source
Score
51.7
Plan availability
Products and plans that support this model
Cursor
Cursor
Cursor is an AI-powered code editor and agentic coding platform by Anysphere. It provides intelligent autocomplete (Tab), autonomous coding agents, cloud agents, code review (Bugbot), and access to frontier models from OpenAI, Anthropic, Google, xAI, and Cursor's own Composer 2 model. Designed for individual developers and engineering teams.
Discussion
Thinking... Make sure you are connected to GitHub server