DeepSeek V3 vs DeepSeek V3.1
| Spec | DeepSeek V3 | DeepSeek V3.1 |
|---|---|---|
| Respan Index | 13.2 (#167) | 18.9 (#137) |
| Blended price per 1M | $0.175 | $0.425 |
| Context window | 128K | 128K |
| Input / output per 1M | $0.14 / $0.28 | $0.25 / $0.95 |
| Max output | 8K | - |
| Input | text | text |
| Released | 2024-12-26 | 2025-08-21 |
Respan Index by category
Each category is scored 0 to 100, and how far ahead the winner is. Leads under a point are close to even.
| Category | DeepSeek V3 | DeepSeek V3.1 |
|---|---|---|
| Coding | 3.1 DeepSeek V3 | 4.6 DeepSeek V3.1 |
| Agents | 5.6* DeepSeek V3 | 10.9 DeepSeek V3.1 |
| Reasoning | 17.3 DeepSeek V3 | 25.6 DeepSeek V3.1 |
| Knowledge & language | 7.9 DeepSeek V3 | 14.6* DeepSeek V3.1 |
| Human preference | 39.4 DeepSeek V3 | 47.8 DeepSeek V3.1 |
* Estimated: no published scores in that category. How the Respan Index works
Benchmarks both report
DeepSeek V3.1 leads on 4 of the 4 benchmarks both report. Higher is better. Each score links to where it was published.
| Benchmark | DeepSeek V3 | DeepSeek V3.1 |
|---|---|---|
| LMArena Elo | 1358 | 1417 |
| SimpleBench (SimpleBench) | 18.9% | 40% |
| WeirdML (Håvard Tveit Ihle) | 36.1% | 38.4% |
| AIME 2025 | 25% | 90.8% |
FAQ
Is DeepSeek V3 better than DeepSeek V3.1?
DeepSeek V3.1 scores 18.9 on the Respan Index (#137), ahead of DeepSeek V3 at 13.2 (#167). DeepSeek V3.1 leads on 4 of the 4 benchmarks both report.
Which is cheaper, DeepSeek V3 or DeepSeek V3.1?
DeepSeek V3 lists at $0.14 input and $0.28 output per 1M tokens; DeepSeek V3.1 at $0.25 and $0.95. DeepSeek V3 is 2.4x cheaper per token at list price.
Which has the larger context window?
Both accept 128K tokens of context.