DeepSeek
DeepSeek V3.1 (Reasoning)
DeepSeek-V3.1 is post-trained on the top of DeepSeek-V3.1-Base, which is built upon the original V3 base checkpoint through a two-phase long context extension approach, following the methodology outlined in the original DeepSeek-V3 report. DeepSeek has expanded their dataset by collecting additional long documents and substantially extending both training phases.
DeepSeek release notesModel Snapshot
TypeLanguage
ReleaseAug 21, 2025
Context163,840
Tagsreasoning, tool-use
Overall
Rank #168 across 526
27.7
Coding
Rank #135 across 436
29.7
Blended Price
Rank #195 across 357
$0.865/M
Strength Profile
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Benchmark Percentiles
Higher bars mean stronger relative placement.
Model Details
Runtime
Max Output8,192
Streaming speed is not measured for this model yet.
Pricing
Blended$0.865/M
Input$0.590/M
Output$1.69/M
Catalog$0.560/M in / $1.68/M out
Metadata
Providers
All Benchmarks
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 27.7 | #168 |
| Artificial Analysis Coding Index | coding | 29.7 | #135 |
| Artificial Analysis Math Index | math | 89.7 | #26 |
| MMLU-Pro | reasoning | 85.1% | #30 |