DeepSeek V3.2-Exp
llmAn experimental update to V3.1-Terminus introducing DeepSeek Sparse Attention for more efficient long-context training and inference at roughly the same output quality.
Metadata
| org | DeepSeek |
| released | 2025-09-29 |
| params | 671B-A37B |
| license | mit |
| kind | llm |
| hf id | deepseek-ai/DeepSeek-V3.2-Exp |
Benchmarks
| MMLU-Pro | 85 | source ↗ |
| GPQA | 79.9 | source ↗ |
| AIME 2025 | 89.3 | source ↗ |
| SWE-bench | 67.8 | source ↗ |