R1 0528
DeepSeekDeepSeek's reasoning-focused "R1" model, built to show its intermediate reasoning steps.
Main uses: Math, logic and multi-step problems where seeing the reasoning trace helps verify the answer.
What sets it apart: A smaller 32,768 max output than DeepSeek V3.2's 147,456 - reasoning traces consume output budget.
In
$0.50
Out
$2.15
Blended
$0.91
Where to buy this model
3 access routes| Route | In | Out | Blended | vs |
|---|---|---|---|---|
| $0.50 | $2.15 | $0.91 | - | |
| ● SiliconFlow fp8 | $0.50 | $2.18 | $0.92 | +1% |
| ● StreamLake | $0.57 | $2.29 | $1.00 | +10% |
● official API · ● cloud platform · ● inference host · USD per 1M tokens. Same model, different providers - price and quantization vary.
- Ctx
- 164K
- max output
- 33K
- cache read /1M
- $0.35
- Family
- deepseek
- Modalities
- text
- Features
- json · reasoning · tools
Live pricing, updated 2026-11-03.
DeepSeek models at a similar price
| Model | In | Out | Ctx |
|---|---|---|---|
| DeepSeek V3.1 | $0.25 | $0.95 | 164K |
| DeepSeek V3 | $0.257 | $1.03 | 164K |
| DeepSeek V3.1 Terminus | $0.27 | $1 | 164K |
| DeepSeek V3 0324 | $0.29 | $1.14 | 164K |
| R1 0528 · this model | $0.5 | $2.15 | 164K |
| DeepSeek V4 Pro 0813 | $0.115 | $4 | 1.048576M |