← All benchmarks
DeepSeek V3.1 (Reasoning)
DeepSeek · Released 2025-08-21 · Independent benchmark scores from Artificial Analysis
Intelligence Index
20.7
Output speed
0.0
tokens/sec
Blended price
$0.86
per 1M tokens
Time to first token
0.00s
Model metadata
| Model ID | 198b717f-42c8-4ab7-a699-ae9373d669d3 |
|---|---|
| Slug | deepseek-v3-1-reasoning |
| Release date | 2025-08-21 |
| Provider | DeepSeek |
| Provider slug | deepseek |
| Output speed | 0.0 tok/s |
| Time to first token | 0.00s |
| Time to first answer | 0.00s |
| Blended price ($/1M) | $0.86 |
| Input price ($/1M) | $0.59 |
| Output price ($/1M) | $1.69 |
All evaluation scores (17)
| aime | — |
|---|---|
| aime 25 | 89.7% |
| artificial analysis coding index | — |
| artificial analysis intelligence index | 20.70 |
| artificial analysis math index | 89.70 |
| gpqa | 77.9% |
| hle | 13.0% |
| ifbench | 41.5% |
| lcr | 53.3% |
| livecodebench | 78.4% |
| math 500 | — |
| mmlu pro | 85.1% |
| scicode | 39.1% |
| tau banking | — |
| tau2 | 37.4% |
| terminalbench hard | 25.0% |
| terminalbench v2 1 | — |
Source: artificialanalysis.ai/models/deepseek-v3-1-reasoning
Benchmark data from Artificial Analysis.