DeepSeek V3.1

Name: DeepSeek V3.1
Brand: DeepSeek
SKU: deepseek/deepseek-v3.1

deepseek/deepseek-v3.1POST /v1/chat/completions

The previous DeepSeek mainline release. Still widely pinned.

Extended reasoningTool callingStructured outputStreaming

Try this model Compare

Input / 1M

$0.2500

via DeepInfra

Output / 1M

$0.9500

via DeepInfra

Cached input / 1M

—

not published

Context

164K

max out 66K

Providers

single route

1 provider serves this model

Multigrid picks between them on every request using your policy. Each price is what that route costs us, and we add no spread of our own to it.

Pin a provider

Provider	Upstream model id	In / 1M	Out / 1M	Billing
DeepInfra	deepseek-ai/DeepSeek-V3.1	$0.2500	$0.9500	Credit

No performance data published yet

Time to first token, throughput and uptime are measurements, and Multigrid has not served enough traffic through DeepSeek V3.1 to publish honest ones. When it has, they will appear here per provider, taken from real requests. Nothing on this page is estimated: prices, context window and capabilities all come from the provider’s own documentation.

Call it

Same request shape as every other model in the catalogue.

example.ts

const res = await client.chat.completions.create({
  model: "deepseek/deepseek-v3.1",
  messages: [{ role: "user", content: "Hello" }],
  reasoning: { effort: "medium" },
});

Chat completions reference Provider preferences

Send DeepSeek V3.1 a real request

Signing up takes no card and commits you to nothing — you start at a zero balance, so nothing can be charged until you decide to add credit. The playground opens on DeepSeek V3.1 and prices every answer as it arrives, at the rate shown above — what the route costs us, with nothing of ours added. Already have an account? The second button skips the form.

Create an account Open the playground