Models·DeepSeek

DeepSeek V4.1 Flash: features and pricing

deepseek/deepseek-v4.1-flash

DeepSeek V4.1 Flash is an open-weight model for text, images, reasoning, and tool use.

$0.30Input, per million tokens
$1.20Output, per million tokens
1.05MContext window
944KLongest answer

1.05M tokens of context — about 786,432 words, on a scale that ends at 2M.

What it can do

  • Reasoning
  • Vision
  • Tool use
  • Open weights
  • Takes text
  • Takes image
  • Gives text

Released 10 Sept 2026 · checked 3 Oct 2026.

About it

DeepSeek V4.1 Flash is a model from DeepSeek with open weights, meaning its model files are available to download. It accepts text and images, can use tools, and supports reasoning. Its context window is 1,048,576 tokens, and it can produce up to 943,718 tokens in one answer. API prices are $0.3 per million input tokens and $1.2 per million output tokens.

Review

In a line

DeepSeek V4.1 Flash is an open-weight AI model from DeepSeek. It can work with text and images, use tools, and perform reasoning.

What it's good at

Its text and image inputs make it suitable for questions that combine written material with pictures. It can also use tools and handle reasoning tasks.

Price in plain terms

API use costs $0.3 per million input tokens and $1.2 per million output tokens. Input is what the model reads; output is what it writes.

At those rates, reading a 500-word page costs about 0.019 cents. A chat turn with about 1,000 tokens in and 500 out costs about 0.09 cents; 1,000 such turns cost about $0.90.

Context and limits

The context window is 1,048,576 tokens, or about how much material the model can consider at once. Its maximum answer is 943,718 tokens. It produces text; no other output type is listed.

Ways to use it

The model has open weights, so its model files are available to download. It is also listed with the model ID deepseek/deepseek-v4.1-flash for use through providers' APIs.

Verdict

Consider it if you need an open-weight model that can take images as well as text, use tools, and reason, with a very large context window. The listed API rates are low for the example workloads provided. It may not fit if you need an output other than text, or if you want a ready-to-use app: no app or chat interface is specified here.

Verdict

DeepSeek V4.1 Flash is worth considering for people comparing API models who need image input, tool use, and reasoning alongside text. Its open weights may also appeal to those who want access to the model files. The listed token rates make small text workloads inexpensive by the provided examples. Look elsewhere if you need audio or other non-text output, or if choosing depends on a particular app or integration; none is specified here.

  • Open weights are available to download.
  • Accepts both text and image input.
  • Supports reasoning and tool use.
  • Large 1,048,576-token context window.
  • Example cost is about $0.90 for 1,000 chat turns.
  • Produces text output only.
  • No app or chat interface is specified.
  • A maximum answer can be large, but no other output type is listed.
  • Pick it ifPeople comparing API costs for text and image tasks; Users who need reasoning or tool use; People looking for an open-weight model
  • Skip it ifYou need audio or another non-text output; You need a named app or integration

Pricing

API pricing is $0.3 per million input tokens and $1.2 per million output tokens. The examples put a page at about 0.019 cents, one chat turn at about 0.09 cents, and 1,000 turns at about $0.90.

Questions

Who makes DeepSeek V4.1 Flash?

DeepSeek makes the model.

Can it read images?

Yes. Its inputs include text and images.

Can it use tools or reason?

Yes. Both tool use and reasoning are listed.

How much can it read at once?

Its context window is 1,048,576 tokens.

How long can one answer be?

The maximum output is 943,718 tokens. The model's listed output type is text.

How much does API use cost?

It costs $0.3 per million input tokens and $1.2 per million output tokens.

Are its model files available?

Yes. It is open-weight.

On these boards

More from DeepSeek