DeepSeek-V3.2-Exp
Zero-eval
#1SimpleQA
#1MMLU-Pro
#1SWE-bench Multilingual
+3 more
by DeepSeek
About
DeepSeek-V3.2-Exp is a language model developed by DeepSeek. It achieves strong performance with an average score of 66.1% across 14 benchmarks. It excels particularly in SimpleQA (97.1%), AIME 2025 (89.3%), MMLU-Pro (85.0%). It supports a 229K token context window for handling large documents. The model is available through 2 API providers. It's licensed for commercial use, making it suitable for enterprise applications. Released in 2025, it represents DeepSeek's latest advancement in AI technology.
Pricing Range
Input (per 1M)$0.27 -$0.27
Output (per 1M)$0.41 -$0.41
Providers2
Timeline
AnnouncedSep 29, 2025
ReleasedSep 29, 2025
License & Family
License
MIT
Performance Overview
Performance metrics and category breakdown
Overall Performance
14 benchmarks
Average Score
66.1%
Best Score
97.1%
High Performers (80%+)
4Performance Metrics
Max Context Window
229.4KAvg Throughput
100.0 tok/sAvg Latency
1msTop Categories
Factuality
97.1%
Math
86.5%
Code
64.3%
General
62.1%
Agents
44.0%
All Benchmark Results for DeepSeek-V3.2-Exp
Complete list of benchmark scores with detailed information
| SimpleQA | factuality | text | 0.97 | 97.1% | Self-reported |
| AIME 2025 | math | text | 0.89 | 89.3% | Self-reported |
| MMLU-Pro | general | text | 0.85 | 85.0% | Self-reported |
| HMMT 2025 | math | text | 0.84 | 83.6% | Self-reported |
| GPQA | general | text | 0.80 | 79.9% | Self-reported |
| Aider-Polyglot | code | text | 0.74 | 74.5% | Self-reported |
| LiveCodeBench | code | text | 0.74 | 74.1% | Self-reported |
| CodeForces | code | text | 0.71 | 70.7% | Self-reported |
| SWE-Bench Verified | general | text | 0.68 | 67.8% | Self-reported |
| SWE-bench Multilingual | general | text | 0.58 | 57.9% | Self-reported |
Showing 1 to 10 of 14 benchmarks