DeepSeek V4 Flash Vision Exp

DeepSeek V4 Flash Vision Exp is a model in the deepseek-flash line, described by the maker as a DeepSeek V4.1 Flash model for reasoning and agentic coding. It is an experimental vision variant, released on 2026-09-10, and follows the previous DeepSeek V4.1 Flash in the same line. It has a context window of 1,000,000 tokens and a maximum output of 393,216 tokens. The weights are open. It supports reasoning mode and tool calling, accepts image and text inputs, and returns text. Pricing starts at $0.0224 per million input tokens and $0.07 per million output tokens, with cached input at $0.0028 per million tokens. It is served by 63 hosts.

First version of its line in the registry.

Range across the hosts that serve it: up to $0.44 per 1M input tokens.

• ·1M token context window
• ·393K max output tokens
• ·Reasoning mode for complex tasks
• ·Tool calling for agentic workflows
• ·Accepts image and text inputs

• Reach for it for reasoning and agentic coding
• Reach for it for long-context analysis with images
• Reach for it for building agents that call tools
• Reach for it for high-volume text generation with large outputs

63 hosts serve this model at the price above · the maker's documentation

The registry keeps the six most recent versions of a line. Older ones stay in the file and can be surfaced without a rebuild.

What is the context window and output limit?

The context window is 1,000,000 tokens and the maximum output is 393,216 tokens.

Does it support image input?

Yes, it accepts image and text inputs, and returns text output.

Is the model open weights and what is the price?

Yes, the weights are open. Input costs from $0.0224 per million tokens, output from $0.07 per million, and cached input from $0.0028 per million.