Model
Model
deepseek
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...
| Provider | Quantization | Input / M tokens | Output / M tokens |
|---|---|---|---|
| DeepInfradeepinfra/fp4 | fp4 | $0.25 | $0.95 |
| SiliconFlowsiliconflow/fp8 | fp8 | $0.27 | $1.00 |
| CoreWeavecoreweave/fp8 | fp8 | $0.55 | $1.65 |
| Maramara | — | $0.60 | $1.70 |
| SambaNovasambanova/fp8 | fp8 | $0.65 | $1.50 |
Prices are what the Provider charges, synced 2026-10-21. Amber adds its platform fee to each Trial’s total; see pricing.
Amber has no curated Model Configuration for this Model yet. Create a custom one in your Workspace to evaluate it.