DeepSeek API pricing
14 models · CN
Open-weight reasoning and coding models priced far below Western equivalents, with off-peak discounts.
- Cheapest input / 1M
- $0.016
- Priciest input / 1M
- $0.700
- Largest context
- 1.0M
- Models tracked
- 14
DeepSeek V4 Flash 0731
R1
DeepSeek V4 Flash 0731
All DeepSeek models
Cheapest first, by blended cost.
| # | Model | Blended / 1M | Input / 1M | Output / 1M | Context | Capabilities |
|---|---|---|---|---|---|---|
| 1 | DeepSeek V4 Flash 0731 DeepSeek | $0.237 | $0.016 | $0.900 | 1.0M | ReasoningToolsOpen weights |
| 2 | DeepSeek V4 Pro 0423 DeepSeek | $0.261 | $0.209 | $0.418 | 1.0M | ReasoningToolsOpen weights |
| 3 | DeepSeek V3.2 Exp DeepSeek | $0.305 | $0.270 | $0.410 | 164K | ReasoningToolsOpen weights |
| 4 | DeepSeek V4 Flash Vision Exp DeepSeek | $0.323 | $0.216 | $0.647 | 1.0M | ReasoningToolsVisionOpen weights |
| 5 | DeepSeek V4 Flash 0423 DeepSeek | $0.343 | $0.030 | $1.28 | 1.0M | ReasoningToolsOpen weights |
| 6 | DeepSeek V3.2 DeepSeek | $0.394 | $0.259 | $0.800 | 164K | ReasoningToolsOpen weights |
| 7 | DeepSeek V3.1 DeepSeek | $0.425 | $0.250 | $0.950 | 164K | ReasoningToolsOpen weights |
| 8 | DeepSeek V3.1 Terminus DeepSeek | $0.453 | $0.270 | $1.00 | 164K | ReasoningToolsOpen weights |
| 9 | DeepSeek V3 DeepSeek | $0.462 | $0.320 | $0.890 | 164K | ToolsOpen weights |
| 10 | DeepSeek V3 0324 DeepSeek | $0.502 | $0.290 | $1.14 | 164K | ToolsOpen weights |
| 11 | DeepSeek V4.1 Flash DeepSeek | $0.525 | $0.300 | $1.20 | 1.0M | ReasoningToolsVisionOpen weights |
| 12 | R1 0528 DeepSeek | $0.913 | $0.500 | $2.15 | 164K | ReasoningToolsOpen weights |
| 13 | DeepSeek V4 Pro 0813 DeepSeek | $0.990 | $0.660 | $1.98 | 1.0M | ReasoningToolsOpen weights |
| 14 | R1 DeepSeek | $1.15 | $0.700 | $2.50 | 64K | ReasoningToolsOpen weights |
Frequently asked
How much does the DeepSeek API cost?
DeepSeek models in this directory range from $0.016 to $0.700 per million input tokens. The cheapest is DeepSeek V4 Flash 0731; the most expensive is R1.
How many models does DeepSeek offer?
14 models are tracked here, with the largest context window being 1.0M tokens on DeepSeek V4 Flash 0731.