AI Rate CardDeepSeek API pricing: V4 Pro and V4 Flash at every provider
Updated Oct 18, 2026
DeepSeek publishes its models with open weights, so besides DeepSeek's own API dozens of hosts and routers sell the same model - at very different prices. The same DeepSeek V4 Pro costs from about $0.16 to $1.74 per million input tokens depending on where you buy it. The tables below are rebuilt from today's prices on every page view.
Two models at a glance
In the 17 October 2026 collection, per 1 million tokens, input / output:
- DeepSeek V4 Pro - $0.66 / $1.98 at DeepSeek itself (also Vercel, Eden AI, OpenCode Go); $1.74 / $3.48 at most US hosts (Azure, Baseten, DigitalOcean, FastRouter). About 60 providers.
- DeepSeek V4 Flash (snapshot 0731) - $0.06 / $0.18 at DeepInfra, Hugging Face and Eden AI, $0.076 / $0.153 at Vercel, around $0.09 / $0.18 at most others. A 1M-token context.
DeepSeek V4 Pro: every provider
| Provider | Model | Input, $ | Output, $ |
|---|---|---|---|
| trustedrouter | DeepSeek: DeepSeek V4 Pro 0423 | $0.16 | $0.63 |
| wafer.ai | DeepSeek V4 Pro | $0.28 | $3.96 |
| tokengo | DeepSeek V4 Pro | $0.43 | $0.87 |
| cometapi | DeepSeek V4 Pro | $0.53 | $1.58 |
| kenari | DeepSeek V4 Pro | $0.55 | $1.10 |
| edenai | DeepSeek V4 Pro | $0.66 | $1.98 |
| orcarouter | DeepSeek V4 Pro | $0.66 | $1.98 |
| zenmux | DeepSeek V4 Pro | $0.66 | $1.98 |
| aiml | DeepSeek V4 Pro | $0.90 | $2.70 |
| aiand | DeepSeek V4 Pro | $1.00 | $2.50 |
| requesty | DeepSeek V4 Pro | $1.04 | $2.08 |
| standardcompute | DeepSeek V4 Pro | $1.04 | $2.09 |
| deepinfra | DeepSeek V4 Pro | $1.30 | $2.60 |
| edenai | DeepSeek V4 Pro | $1.30 | $2.60 |
| fastrouter | DeepSeek: DeepSeek V4 Pro | $1.30 | $2.60 |
Live data: this table is built from today's prices.
Why the same model costs 2.6 times more at an American cloud than at DeepSeek: you pay for where the request is served (US or EU data centres, the provider's own privacy terms and limits), not for a different model. If your data may go to DeepSeek's own servers, its API is the cheapest full-price option; if it must stay with a US or EU host, compare the $1.74 group by output price and limits. A few routers list prices far below DeepSeek's own (trustedrouter $0.16 / $0.63) - check their terms and the date on the card before you rely on them.
DeepSeek V4 Flash 0731: every provider
| Provider | Model | Input, $ | Output, $ |
|---|---|---|---|
| trustedrouter | DeepSeek: DeepSeek V4 Flash 0731 | $0.0422 | $0.09495 |
| openrouter | DeepSeek: DeepSeek V4 Flash 0731 | $0.059 | $0.30 |
| deepinfra | DeepSeek V4 Flash 0731 | $0.06 | $0.18 |
| edenai | DeepSeek V4 Flash 0731 | $0.06 | $0.18 |
| huggingface | DeepSeek V4 Flash 0731 | $0.06 | $0.18 |
| requesty | DeepSeek V4 Flash 0731 | $0.072 | $0.14 |
| requesty | DeepSeek V4 Flash 0731 | $0.076 | $0.15 |
| cortecs | DeepSeek V4 Flash 0731 | $0.08272 | $0.41 |
| meganova | DeepSeek-V4-Flash-0731 | $0.09 | $0.18 |
| requesty | DeepSeek V4 Flash 0731 | $0.09 | $0.18 |
| requesty | DeepSeek V4 Flash 0731 | $0.09 | $0.18 |
| vultr | DeepSeek V4 Flash 0731 | $0.10 | $0.25 |
| inferx | DeepSeek V4 Flash 0731 | $0.11 | $0.22 |
| melious | DeepSeek V4 Flash 0731 | $0.11 | $0.28 |
| edenai | DeepSeek V4 Flash 0731 | $0.12 | $0.24 |
Live data: this table is built from today's prices.
The cheapest row is routing.run at $0.03 / $0.06; the established hosts start at $0.06 / $0.18. Flex tiers (Requesty, Neuralwatt) are a little cheaper because a request may wait.
DeepSeek V4 Flash: every provider
| Provider | Model | Input, $ | Output, $ |
|---|---|---|---|
| orcarouter | DeepSeek V4 Flash (free) | Free | Free |
| standardcompute | DeepSeek V4 Flash | $0.0826 | $0.17 |
| deepinfra | DeepSeek V4 Flash | $0.09 | $0.18 |
| edenai | DeepSeek V4 Flash | $0.09 | $0.18 |
| fastrouter | DeepSeek: V4 Flash | $0.09 | $0.18 |
| huggingface | DeepSeek V4 Flash | $0.09 | $0.18 |
| kilo | DeepSeek: DeepSeek V4 Flash 0423 | $0.091 | $0.18 |
| neuralwatt | DeepSeek V4 Flash Flex | $0.091 | $0.18 |
| openrouter | DeepSeek: DeepSeek V4 Flash 0423 | $0.091 | $0.18 |
| trustedrouter | DeepSeek: DeepSeek V4 Flash 0423 | $0.09495 | $0.19 |
| meganova | DeepSeek-V4-Flash | $0.10 | $0.20 |
| tokengo | DeepSeek V4 Flash | $0.10 | $0.20 |
| cometapi | DeepSeek V4 Flash | $0.12 | $0.48 |
| vercel | DeepSeek V4 Flash | $0.13 | $0.26 |
| digitalocean | DeepSeek V4 Flash | $0.14 | $0.28 |
Live data: this table is built from today's prices.
OpenCode and OrcaRouter list a free "Flash Free" variant - fine for trying the model, but free tiers come with tight rate limits and may log your prompts.
How to choose
- Cheapest strong model for coding and reasoning: V4 Pro at DeepSeek's own API.
- Data must stay in the US or EU: V4 Pro at a US host - pick by output price and limits.
- Chatbots, extraction, high volume: V4 Flash - an output price a tenth of Pro's.
- Just trying it: the free Flash variants. To estimate a month, put your own volume into the cost calculator.
All prices are in USD per 1 million tokens, without taxes, as the providers publish them; each row links to its card with the source and the date it was checked.