Models·DeepSeek

DeepSeek V4.1 Flash (batch)

deepseek/deepseek-v4.1-flash:batch

DeepSeek V4.1 Flash (batch) is an open-weight text model that handles images, reasoning, and tools.

$0.11Input, per million tokens
$0.34Output, per million tokens
1.05MContext window
131KLongest 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 (batch) is an open-weight model from DeepSeek. It accepts text and images, produces text, supports reasoning and tools, and can read up to 1,048,576 tokens at once. Input costs $0.112 per million tokens and output costs $0.336 per million tokens. It was released on September 10, 2026.

Review

In a line

DeepSeek V4.1 Flash (batch) is a DeepSeek model built as a sparse mixture of experts. It is the first DeepSeek model based on the company’s Causal Encoder-Decoder architecture.

It has open weights, accepts text and images, and returns text. See other DeepSeek models for related entries.

What it's good at

The model supports reasoning, tool use, and image input. That makes it suited to tasks that combine written instructions with visual information or require step-by-step answers.

Its open weights may suit people looking for open-weight models.

Price in plain terms

Input costs $0.112 per million tokens. Output costs $0.336 per million tokens.

Reading a page of about 500 words costs under 0.01 cents. A chat turn with about 1,000 tokens in and 500 out costs about 0.028 cents. At that same size, 1,000 turns cost about $0.28.

These are token-use charges, so longer prompts and answers use more of the paid input or output total.

Context and limits

The context window is 1,048,576 tokens, meaning it can take in a very large amount of text or other input in one request.

One answer can contain up to 131,072 tokens. The model outputs text only, even though it can accept images.

Ways to use it

Use the model ID deepseek/deepseek-v4.1-flash:batch wherever a service offers this batch model. Send text or images, and receive a text response.

Its open weights also make it relevant to people comparing newest models and models they can examine or run themselves.

Verdict

Choose this model if you need a very large input limit, image understanding, reasoning, and tool support at a low token price. Its open weights may also matter when control over the model is important.

Look elsewhere if you need it to produce images or another output type, because its output is text. Also check that the service you use offers the batch model ID before planning around it.

Verdict

DeepSeek V4.1 Flash (batch) is a practical option for high-volume text work where image input, reasoning, or tool calls matter. The token rates are low enough for frequent use, including about $0.28 for 1,000 medium-sized chat turns. Its million-token context window fits very large documents, while the 131,072-token output ceiling supports long replies. It is not a fit for image generation or other non-text results, and availability depends on a service supporting the batch model identifier.

  • Handles text and image input
  • Supports reasoning and tool use
  • Open weights are available
  • Context window reaches 1,048,576 tokens
  • Low listed input and output rates
  • Produces text only
  • Batch model availability may vary by service
  • Output is capped at 131,072 tokens
  • Pick it ifLarge-document text tasks; Image-aware reasoning and tool workflows; Buyers comparing low token costs
  • Skip it ifYou need image generation or non-text output; Your provider does not offer the batch model ID

Pricing

Input is $0.112 per million tokens and output is $0.336 per million tokens. A typical 1,000-token-in, 500-token-out turn costs about 0.028 cents; 1,000 such turns cost about $0.28.

Questions

Who makes DeepSeek V4.1 Flash (batch)?

DeepSeek makes it. The model belongs to the deepseek family.

Can it understand images?

Yes. Images are supported as input, and the model returns text.

Does it support reasoning and tools?

Yes. The listed capabilities include reasoning and tool use.

How much context can it read?

Its context window is 1,048,576 tokens.

How long can one answer be?

The maximum output is 131,072 tokens.

What does it cost?

Input costs $0.112 per million tokens and output costs $0.336 per million tokens.

Is it open weight?

Yes. The facts list open weights as available.

What does it output?

It outputs text only.

On these boards

More from DeepSeek