Claude Sonnet 4.6
Parameters
Undisclosed
Context Length
1M
Modality
Multimodal
Architecture
Undisclosed
License
Proprietary
Release Date
17 Feb 2026
Knowledge Cutoff
Aug 2025
API Pricing (per 1M)
Input: $3.00 · Output: $15.00
Evaluation Benchmarks
Rank
#38
| Benchmark | Score | Rank |
|---|---|---|
Coding | medium 0.79 | 17 |
Data Analysis | medium 0.78 | 18 |
Graduate-Level QA | 0.899 | 20 |
Reasoning | medium 0.85 | 28 |
General Text | 1473 | 29 |
Web Development | 1521 | 32 |
Agent Arena | -1.27 | 33 |
General | medium 0.73 | 34 |
LiveBench Average | medium 0.73 | 34 |
Mathematics | medium 0.87 | 35 |
Agentic Coding | medium 0.43 | 39 |
Agentic Index | max 0.33 | 49 |
Coding Index | max 0.63 | 58 |
Intelligence Index | max 0.30 Standard 0.25 | 69 95 |
Rankings
Overall Rank
#38
Coding Rank
#40
About Claude Sonnet 4.6
Claude Sonnet 4.6 is a high-performance multimodal workhorse from Anthropic optimized for agentic workflows, GUI navigation, and complex codebases. It features hybrid adaptive reasoning and context compaction across a 1M token context window.
Technical Specifications
Architecture specifications are undisclosed for proprietary models.
Attention
Attention Structure
Multi-Head Attention
Attention Heads
Undisclosed
Key-Value Heads
Undisclosed
Attention Head Dimension
Undisclosed
Position Embedding
Absolute Position Embedding
RoPE Theta
Undisclosed
Sliding Window Attention
Undisclosed
Sliding Window Size
Undisclosed
Sliding Window Ratio
Undisclosed
Linear Attention
Undisclosed
Linear Attention Ratio
Undisclosed
Normalization
Undisclosed
Activation Function
Undisclosed
Dimensions
Hidden Dimension Size
4,096
Number of Layers
Undisclosed
FFN Intermediate Size (Dense)
Undisclosed
Multi-Token Prediction Heads
Undisclosed
Tokenizer
Vocabulary Size
Undisclosed
Resources
About Claude 4
Anthropic's fourth generation Claude models with advanced reasoning, extended context windows up to 200K tokens, and configurable thinking effort levels. Features improved safety alignment, nuanced understanding, and sophisticated task completion. Includes Opus (most capable), Sonnet (balanced), and Haiku (fast) variants, with thinking modes that enable transparent chain-of-thought reasoning for complex problems.