Up to four models
DeepSeek-V3.2 vs DeepSeek-V3.2-Exp
On the index (GPT-4 = 100), DeepSeek-V3.2 scores 141 and DeepSeek-V3.2-Exp 140. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
141125–157
WeightsOpen
ReleasedDec 2025
Price in / out$0.26 / $0.38
Context164K
Core tests taken2 of 9
140124–157
WeightsOpen
ReleasedSep 2025
Price in / out$0.27 / $0.41
Context164K
Core tests taken0 of 9
DeepSeek-V3.2 and DeepSeek-V3.2-Exp are too close to call (1 point apart, within the uncertainty), at a similar price.
Test by testthird-party results
Every result as a table
| Test | DeepSeek-V3.2 | DeepSeek-V3.2-Exp |
|---|---|---|
| Reasoning & knowledge | ||
| CL-bench | 12.4% | 13.2% |
| Coding | ||
| Aider Polyglot | 74.2% | 74.2% |
| Agents | ||
| DTBench | 85.6% | 87.7% |
| LMCA | 28.8% | 29.1% |
Tests at least two of these models have taken. Bold rows are the core tests.