DeepSeek models: prices and the hardware each needs
30 models, 7 sold per token by 13 vendors. Vendors are in alphabetical order; the price is the cheapest list price among them, per million tokens.
| Model | Size | Sold per token by | From, input | Output | GPU memory |
|---|---|---|---|---|---|
| DeepSeek V3.1 | 671 B | AWS, DeepInfra | $0.25 | $0.95 | 855.7 GB |
| DeepSeek V3.2 | 685 B | AWS, Azure, DeepInfra | $0.26 | $0.38 | 856.7 GB |
| DeepSeek V4 Flash | 291 B | Azure, Databricks, DeepInfra, Huawei Cloud, Snowflake | $0.09 | $0.18 | 199.5 GB |
| DeepSeek V4 Flash 0731 | 304 B | Cloudflare, Crusoe, DeepInfra, Nebius | $0.06 | $0.18 | 208.6 GB |
| DeepSeek V4 Pro | 1599 B | Alibaba Cloud, Azure, Core42, Crusoe, Databricks, DeepInfra, Nebius | $1.30 | $2.60 | 1080.9 GB |
| DeepSeek V4 Pro 0813 | 1650 B | Cloudflare, DeepInfra, Nebius, Together AI | $1.30 | $2.60 | 1115.9 GB |
| DeepSeek V4.1 Flash | 763 B | Alibaba Cloud, Azure, DeepInfra, Fireworks AI, Together AI | $0.14 | $0.42 | 637.9 GB |
| deepseek-coder-1.3b-instruct | 1.35 B | not sold per token | 3.4 GB | ||
| deepseek-coder-6.7b-base | 6.74 B | not sold per token | 16.9 GB | ||
| deepseek-coder-6.7b-instruct | 6.74 B | not sold per token | 16.9 GB | ||
| deepseek-coder-7b-instruct-v1.5 | 6.91 B | not sold per token | 17.3 GB | ||
| DeepSeek-Coder-V2-Lite-Instruct | 15.71 B | not sold per token | 39.3 GB | ||
| deepseek-moe-16b-chat | 16.38 B | not sold per token | 40.9 GB | ||
| DeepSeek-R1 | 684.49 B | not sold per token | 855.6 GB | ||
| DeepSeek-R1-0528 | 684.53 B | not sold per token | 855.7 GB | ||
| DeepSeek-R1-0528-Qwen3-8B | 8.19 B | not sold per token | 20.5 GB | ||
| DeepSeek-R1-Distill-Llama-70B | 70.55 B | not sold per token | 176.4 GB | ||
| DeepSeek-R1-Distill-Llama-8B | 8.03 B | not sold per token | 20.1 GB | ||
| DeepSeek-R1-Distill-Qwen-1.5B | 1.78 B | not sold per token | 4.5 GB | ||
| DeepSeek-R1-Distill-Qwen-14B | 14.77 B | not sold per token | 36.9 GB | ||
| DeepSeek-R1-Distill-Qwen-32B | 32.76 B | not sold per token | 81.9 GB | ||
| DeepSeek-R1-Distill-Qwen-7B | 7.62 B | not sold per token | 19.1 GB | ||
| DeepSeek-V2 | 235.74 B | not sold per token | 589.4 GB | ||
| DeepSeek-V2-Chat | 235.74 B | not sold per token | 589.4 GB | ||
| DeepSeek-V2-Lite | 15.71 B | not sold per token | 39.3 GB | ||
| DeepSeek-V2-Lite-Chat | 15.71 B | not sold per token | 39.3 GB | ||
| DeepSeek-V3-0324 | 684.53 B | not sold per token | 855.7 GB | ||
| DeepSeek-V3.1-Base | 684.53 B | not sold per token | 855.7 GB | ||
| DeepSeek-V3.1-Terminus | 684.53 B | not sold per token | 855.7 GB | ||
| DeepSeek-V3.2-Exp | 685.4 B | not sold per token | 856.8 GB |