DeepSeek › DeepSeek V4 Pro 0813 › Pricing
DeepSeek V4 Pro 0813 pricing
API list price by DeepSeek · checked Oct 23, 2026
$1.32 / $3.96per 1M tokens in / out
$0.044cache read · −97%
$0.36per task at non-thinking?
#29on OpenRouter by tokens · 988B / week
Price sheetper 1M tokens
| Input | $1.32 | list |
| Output | $3.96 | |
| Cache read | $0.044 | −97% vs input |
| Context · max output | 1M · 384K | tokens |
Cheapest host native quote has input above output (fp4). Components are retained as advertised.
Cost per task by effort?
non-thinkingAPI default$0.3610.6
max$0.6741
per taskHumanity's Last Exam
Same price per token at every effort; the bill grows with reasoning tokens. max costs 1.9× non-thinking for +30.4 points.
Where to buy19 providers · per 1M tokens
| Provider | Input | Output | Context |
|---|---|---|---|
| Wafer | $0.25 | $5 | 1M |
| Ionstream | $0.37 | $2.93 | 1M |
| Alibaba | $0.58 | $1.74 | 1M |
| DekaLLM | $0.88 | $3.80 | 1M |
| Sail Research | $0.90 | $0.58 | 1M |
| Relace | $0.90 | $1.50 | 1M |
| DigitalOcean | $1.04 | $2.09 | 1M |
| DeepInfra | $1.30 | $2.60 | — |
| CoreWeave | $1.31 | $3.96 | 1M |
| AtlasCloud | $1.32 | $3.96 | 1M |
| Cloudflare | $1.32 | $3.96 | 1M |
| deepseeklab | $1.32 | $3.96 | — |
| Novita | $1.32 | $3.96 | — |
| Parasail | $1.32 | $3.96 | 1M |
| Together | $1.32 | $3.96 | 1M |
| SiliconFlow | $1.50 | $3.14 | 1M |
| Venice | $1.65 | $3.30 | 1M |
| Azure | $1.91 | $3.83 | 1M |
| Azure AI Foundry | $1.93 | $3.83 | — |
In DeepSeek's line-upper 1M in / out
| DeepSeek V4.1 Flash 1017 | $0.30 / $1.20 | −73% |
| DeepSeek R1 Distill Qwen 14B | $0.43 / $1.12 | −70% |
| R1 0528 | $0.39 / $1.63 | −65% |
| DeepSeek R1 Distill Qwen 7B | $0.68 / $0.80 | −64% |
| DeepSeek R1 Distill Llama 70B | $0.70 / $0.80 | −63% |
| DeepSeek R1 | $0.70 / $2.40 | −43% |
| DeepSeek V4 Pro 0813 | $1.32 / $3.96 | this model |
Cheaper for the same workvs DeepSeek V4 Pro 0813
| GPT-6.1 Sol high | -10% | score ×2.5 | $0.32 |
| MiMo v2.6 Pro thinkingOpen | -63% | score ×2.3 | $0.13 |
| DeepSeek V4.1 Flash 1017 maxOpen | -26% | score ×2.2 | $0.26 |
| Claude Haiku 5.5 max | -40% | score ×2.1 | $0.21 |
| MiniMax M3.1 Flash thinking | -38% | score ×2.1 | $0.22 |
Pricing questions
How much does DeepSeek V4 Pro 0813 cost?›
$1.32 per 1M input tokens and $3.96 per 1M output tokens. Cached input is $0.044. One Intelligence Index task costs about $0.36 at the default effort.
Is DeepSeek V4 Pro 0813 free to use?›
There is no free API tier, but the weights are open: you can run it on your own hardware for the cost of compute.
How much does prompt caching save?›
Cache reads cost $0.044 per 1M, 97% below the input price.
Is DeepSeek V4 Pro 0813 cheaper than DeepSeek V4 Pro 0424?›
No: $1.32 / $3.96 against $0.43 / $0.87, 264% more per token.
How does effort change the price?›
The price per token is the same at every effort. Higher effort spends more reasoning tokens: a task costs $0.36 at non-thinking, $0.67 at max.
What is a cheaper alternative to DeepSeek V4 Pro 0813?›
GPT-6.1 Sol at high effort costs -10% per task with a score ×2.5 on Humanity's Last Exam.