DeepSeek V3.2 (Non-reasoning)

Available
DeepSeek logo
DeepSeek
2025-12-01
32,000 tokens

DeepSeek 开发的 AI 模型,适合多种 AI 应用场景。

Supported modalities:TextCode
Quick Overview
Text Generation3/10
Code Generation3/10
Reasoning6/10
Multimodal3/10
Benchmark Results
Scores from leading benchmark suites.
32.1%
artificial-analysis-intelligence
34.6%
artificial-analysis-coding
59.0%
artificial-analysis-math
Performance Metrics
Latency and throughput performance.
1000ms
P50 Latency
31.036
tokens/sec
Pricing Info
Input (1K tokens)$0.00028000000000000003
Output (1K tokens)$0.00041999999999999996
View detailed pricing
Feature support
Streaming
Tool calling
JSON mode
DeepSeek V3.2 (Non-reasoning) Model Review & Specs