·
DeepSeek-V3.1 is a large hybrid reasoning model (671B parameters, 37B active) that supports both thinking and non-thinking modes via prompt templates. It extends the DeepSeek-V3 base with a two-phase long-context...
| 信号 | 强度 | 权重 | 影响 |
|---|---|---|---|
| Benchmarksjust now | 69 | 30% | +20.7 |
| Pricingjust now | 99 | 15% | +14.9 |
| Capabilitiesjust now | 67 | 20% | +13.3 |
| Recencyjust now | 85 | 15% | +12.8 |
| Context Windowjust now | 72 | 10% | +7.2 |
| Output Capacityjust now | 64 | 10% | +6.4 |
LMSYS Arena Elo
1418
Percentile
86.3
Weight
30%
No task benchmark data available yet for this model.
把当前模型放回同一服务商最近的发布节奏中查看。
社区和从业者反馈在基准测试和价格之上增加了真实世界的信号。
Share your experience with DeepSeek V3.1 and help the community make better decisions.
LMSYS Arena Elo
1418
Percentile
86.3
Weight
30%
No task benchmark data available yet for this model.
成本估算器
每月比类别平均节省$42.34