DeepSeek V4.1 Flash (ModelArk)

Published rate

Input

$0.1875 / 1M

Cached input

$0.00375 / 1M

Output

$0.75 / 1M

Context

1,048,576 tokens

Max output

393,216 tokens

Verified against the live registry on 2026-11-04. If this page and /v1/models disagree, the API is correct and this page is a bug.

Routing and retention

Model type

Third-party model, served through Fikra

Upstream family

deepseek

Provider

byteplus

Upstream ID

deepseek-v4-1-flash-261017

Retention

Prompts are not used for training. Response bodies may be cached for up to 300 seconds for non-ZDR requests. A ZDR request disables that cache. Upstream retention varies by provider; the model page lists the provider.

Call it

OpenAI-compatible. One line changes.

from openai import OpenAI client = OpenAI( base_url="https://api.fikraapi.co.ke/v1", api_key="fk_..." ) response = client.chat.completions.create( model="deepseek-v4.1-flash-ark", messages=[{"role":"user","content":"Hello"}] )

Try it first

A playground is available for chat models. The quote endpoint returns a price before you run a call.

Open playground

Quote endpoint ->

Need a key? Start with $0.5 credit.