DeepSeek V4 Pro

DeepSeek V4 Pro is a snapshot in the deepseek-thinking line, released on 2026-08-12. It supports thinking and non-thinking modes and provides a million-token context window, as described by the maker. The model has a context window of 1,000,000 tokens and a maximum output of 393,216 tokens. It accepts text and returns text, with tool calling and reasoning mode enabled. Weights are open. It is served by 65 hosts. Pricing starts at $0.2088 per million input tokens and $0.4176 per million output tokens, with cached input from $0.003 per million tokens.

First version of its line in the registry.

Range across the hosts that serve it: up to $2.40 per 1M input tokens.

• ·Million-token context window
• ·Supports thinking and non-thinking modes
• ·Tool calling enabled
• ·Open weights for self-hosting
• ·Output up to 393,216 tokens

• Reach for it for long-document analysis
• Reach for it for multi-step reasoning with thinking mode
• Reach for it for tasks needing tool calls
• Reach for it for self-hosted deployments

65 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?

The context window is 1,000,000 tokens.

Does it support tool calling?

Yes, tool calling is supported.

Are the weights open?

Yes, the weights are open.