AIbase
EN
Home
AI Models Finder
Claude 4 Sonnet (Extended Thinking)
C

Claude 4 Sonnet (Extended Thinking)

Balanced version of Claude 4 with extended thinking functionality, providing visible reasoning processes. Achieves 72.7% on SWE-bench (80.2% with parallel compute), offers better cost-effectiveness compared to Opus. Supports powerful coding and analysis capabilities with complex multi-step reasoning, ideal for professional applications requiring deep thinking with cost considerations.
Intelligence(Medium)
Speed(Relatively Slow)
Input Supported Modalities
Yes
Is Reasoning Model
200,000
Context Window
128,000
Maximum Output Tokens
-
Knowledge Cutoff
Go Compare

Pricing

￥21.6 /M tokens
Input
￥108 /M tokens
Output
￥43.2 /M tokens
Blended Price

Quick Simple Comparison

Input

Output

Claude Opus 4.1
￥15
Claude 4 Sonnet
￥3
Claude 3.5 Haiku
￥0.8

Basic Parameters

Claude 4 Sonnet (Extended Thinking)Technical Parameters
Parameter Count
Not Announced
Context Length
200.00k tokens
Training Data Cutoff
Open Source Category
Proprietary
Multimodal Support
Text, Image
Throughput
101
Release Date
2025-05-22
Response Speed
85.094,666 tokens/s

Benchmark Scores

Below is the performance of Claude 4 Sonnet (Extended Thinking) in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.
Intelligence Index
60.78
Large Language Model Intelligence Level
Coding Index
49.02
Indicator of AI model performance on coding tasks
Math Index
83.93
Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks
MMLU Pro
84.6
Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video
GPQA
72.1
Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions
HLE
8.5
The model's comprehensive average score on the Hugging Face Open LLM Leaderboard
LiveCodeBench
58
Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems
SciCode
40
The model's capability in code generation for scientific computing or specific scientific domains
HumanEval
-
Score achieved by the AI model on the specific HumanEval benchmark test set
Math 500 Score
97.5
Score on the first 500 larger, more well-known mathematical benchmark tests
AIME Score
70.3
An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase