DeepSeek-V4-Pro

Name: DeepSeek-V4-Pro
Author: DeepSeek

by DeepSeek · China

Open weightsTool callingMixture of experts1M context

The open-weight flagship of the DeepSeek V4 preview, released with V4-Flash as DeepSeek's first million-token models. DeepSeek claimed open-source state of the art on agentic coding but published no score on the benchmarks tracked here.

Specification

The numbers

Maker: DeepSeek
Released: 2026-04
Parameters: 1.6T total / 49B active
Architecture: MoE
Context window: 1,048,576 tokens
Max output: Not reported
Input: Text
Output: Text
Reasoning: Not reported
Tool calling: Yes
Knowledge cutoff: Not reported
Licence: MIT
Availability: Not reported
Weights: Downloadable

Cost

Price per million tokens

No public API price. The weights are published, so the only cost is the hardware you run it on.

Published scores

Benchmarks

Figures published by DeepSeek or taken from a public leaderboard. Row last checked 2026-10.

Every reported test

No published benchmark scores yetIts maker has not published benchmark figures for this model, and nothing has been estimated in their place.

Where these numbers come from

Every score on this page is a published figure, taken from the model's own card, system card, technical report or release post, or from a public leaderboard. CorX Labs did not run these evaluations. Most are self-reported by the lab that built the model, which means they were produced under that lab's own choice of prompt, scaffold and number of attempts — so treat them as a starting point for a shortlist, not as a settled ranking.

A score someone other than the model's maker measured is marked Independent and names its measurer. Those are the stronger numbers on this page — an outside harness has no reason to flatter anyone — and there are not many of them.

Where a figure has not been published, the cell reads Not reported rather than an estimate. Nothing here is inferred, interpolated or guessed. Each model records the month its row was last checked. Full method and caveats.

Head to head

DeepSeek-V4-Pro compared

vs DeepSeek-V4.1-FlashDeepSeek vs DeepSeek-V3.2DeepSeek vs Mistral Large 4Mistral AI

Compare with something else

Same maker

Other models from DeepSeek

AI models with context window, price per million tokens and published benchmark scores. Sortable by any column.

DeepSeek-V4.1-FlashDeepSeek	DeepSeek	1M	$0.30	$1.20	—	90.9%	—	—	Not stated
DeepSeek-V3.2DeepSeek	DeepSeek	164K	$0.280	$0.42	85%	79.9%	89.3%	—	MIT
DeepSeek-V3.1DeepSeek	DeepSeek	131K	$0.280	$1.14	84.8%	80.1%	88.4%	66%	MIT
DeepSeek-R1DeepSeek	DeepSeek	131K	$0.550	$2.19	84%	71.5%	79.8%	49.2%	MIT
DeepSeek-R1-Distill-Qwen-32BDeepSeek	DeepSeek	131K	$0.12	$0.18	—	62.1%	72.6%	—	MIT
DeepSeek-R1-Distill-Llama-8BDeepSeek	DeepSeek	131K	$0.04	$0.04	—	—	50.4%	—	MIT
DeepSeek-V3DeepSeek	DeepSeek	131K	$0.27	$1.10	75.9%	59.1%	—	—	MIT
DeepSeek-Coder-V2DeepSeek	DeepSeek	131K	$0.140	$0.280	60%	—	—	—	DeepSeek License

DeepSeek