The Model Index
Models/DeepSeek

DeepSeek-V4-Pro

DeepSeekOpen weightsFlagship

DeepSeek-V4-Pro is an open-weights AI model from DeepSeek with an index score of 163 (GPT-4 = 100), rank 40 of 217.

163147–178Index score#40 of 217 models
$0.11/ $4.44Price per million tokensinput / output · via OpenRouter
1.05MContext windowtokens it reads at once
13 Aug 2026Released3 months ago

Where it sitson the frontier

801001201401601802002023202420252026GPT-4 at launch = 100DeepSeek-V4-Pro

Claude Opus 4.6 reached this score 6 months earlier.

Test resultsby area · tick = best by any model

Reasoning & knowledgeevidence
GPQA Diamond (Epoch AI)
92%
Humanity's Last Exam
no result
Mathsevidence
FrontierMath (tiers 1-3)
65%
Codingevidence
SWE-bench Verified
no result
Terminal-Bench 4.0
14%
Agentsevidence
APEX-Agents
47%
Professional workevidence
Finance Agent (v2)
50%
Novel problemsevidence
ARC-AGI-2
61%
Long tasksno evidence yet
METR time horizon
no result

Third-party results only, at the model’s best standard setting. Hover a bar for who ran it.

What others sayrank on 2 leaderboards

The Model Indexthird-party tests, our method#40of 217 · 163
Epoch Capabilities Indexoverall capability from many benchmarks#41of 231 · 156.0

Best attop three among current models

Not in the top three among current models on any category we track.

Where to use itprices checked 18 Nov 2026

Host prices from OpenRouter; going direct can differ.

Facts and sources

Released13 Aug 2026
Release entryDeepSeek-V4-Pro (GA)
Scored asDeepSeek V4 Pro 0813
Output speed–

DeepSeek releasesaround this one

This page as a chartDownload PNG