DeepSeek V4.1 Flash API prices by provider
Last updated: 8 Oct 2026
The short answer
On a coding-agent workload, DeepSeek V4.1 Flash costs $0.0110 per 1M tokens on OpenScale. DeepInfra is next at $0.0218, so OpenScale costs 49% less. The median of the 12 other providers we checked is $0.0414.
Price per 1M tokens, cheapest first
Input, cached input and output are each provider's public price in USD per 1M tokens. Effective is the price of 1M tokens of a coding-agent workload: 99% input tokens and 1% output, with 90% of input tokens served from cache.
| Provider | Input | Cached | Output | Effective |
|---|---|---|---|---|
| OpenScale | $0.06 | $0.003 | $0.24 | $0.0110 |
| DeepInfra | $0.14 | $0.0042 | $0.42 | $0.0218 |
| Relace | $0.0346 | $0.01 | $1.00 | $0.0223 |
| DeepSeek | $0.15 | $0.003 | $0.60 | $0.0235 |
| StreamLake | $0.15 | $0.003 | $0.60 | $0.0235 |
| NovitaAI | $0.195 | $0.0039 | $0.78 | $0.0306 |
| DigitalOcean | $0.225 | $0.006 | $0.90 | $0.0366 |
| Makora | $0.27 | $0.009 | $1.15 | $0.0462 |
| Parasail | $0.30 | $0.006 | $1.20 | $0.0470 |
| Together | $0.30 | $0.006 | $1.20 | $0.0470 |
| Venice | $0.30 | $0.0075 | $1.20 | $0.0484 |
| Wafer | $0.045 | $0.044 | $1.20 | $0.0557 |
| AtlasCloud | $0.30 | $0.03 | $1.20 | $0.0684 |
What a month of agent traffic costs
A team whose coding agents send 1B tokens a month, on the same workload:
- OpenScale: $11.01 a month
- DeepInfra, the next cheapest: $21.80 a month
- The median provider: $41.44 a month
DeepSeek V4.1 Flash on OpenScale
| Model id | deepseek/deepseek-v4.1-flash |
|---|---|
| Made by | DeepSeek |
| Context window | 1,048,568 tokens |
| Max output | 384,000 tokens |
| Input | text and image |
| Reasoning effort | low, high, max |
| Tool calling | Yes |
| Structured outputs | Yes |
| Versions | 1017 (Oct 17, 2026), 0910 (Sep 10, 2026) |
The plain id always runs the latest version we have tested. A pinned id, such as deepseek/deepseek-v4.1-flash-1017, never changes.
How we compare
The table lists the providers that serve a real share of public DeepSeek V4.1 Flash traffic, at their public price after any discount, checked in 8 Oct 2026. Measured cache hit rates swing from day to day, so the effective price uses one workload for every provider, OpenScale included: 99% input tokens and 1% output, with 90% of input tokens served from cache. Providers whose measured cache hit rate is far below that are left out, because the shared rate would make them look cheaper than they are.
Questions
- Which API provider is cheapest for DeepSeek V4.1 Flash?
- OpenScale, at $0.0110 per 1M tokens on a coding-agent workload (99% input tokens and 1% output, with 90% of input tokens served from cache). The next cheapest we checked is DeepInfra at $0.0218.
- How much does DeepSeek V4.1 Flash cost on OpenScale?
- $0.06 per 1M input tokens, $0.003 per 1M cached input tokens and $0.24 per 1M output tokens, billed by usage with no monthly fee.
- Why rank by an agent workload and not by the input price?
- A coding agent sends its whole conversation again on every step, so almost all of its tokens are input, and most of that input was already sent before and is served from cache. The cached input price decides most of the bill, so a low input price alone can still cost more.
- Can I call DeepSeek V4.1 Flash on OpenScale with the OpenAI or Anthropic SDK?
- Yes. OpenScale serves the OpenAI Chat Completions and Responses APIs and the Anthropic Messages API. Point an existing SDK or coding agent at OpenScale with your API key and the model id deepseek/deepseek-v4.1-flash.
- How current are these prices?
- Other providers' public prices were checked in 8 Oct 2026, after any discount they show. OpenScale prices come from our live price list.
Start building
Create an account to get an API key, then set up your coding agent or read the API reference. Every comparison is on the compare page.