DeepSeek-V3 (Mar 2025)
DeepSeek-V3 (Mar 2025) is an open-weights AI model from DeepSeek with an index score of 124 (GPT-4 = 100), rank 147 of 217.
124108–140Index score#147 of 217 models
$0.29/ $1.14Price per million tokensinput / output · via OpenRouter
164KContext windowtokens it reads at once
24 Mar 2025ReleasedReplaced by DeepSeek-V3.1
Where it sitson the frontier
o1 reached this score 4 months earlier.
Test resultsby area · tick = best by any model
Reasoning & knowledgeevidence
Mathsevidence
Codingevidence
Agentsevidence
Professional workno evidence yet
Novel problemsno evidence yet
Long tasksevidence
Third-party results only, at the model’s best standard setting. Hover a bar for who ran it.
What others sayrank on each leaderboard
The Model Indexthird-party tests, our method#147of 217 · 124
Best attop three among current models
Replaced by DeepSeek-V3.1; it no longer leads any category among current models.
Where to use itprices checked 18 Nov 2026
API providers 2 · per 1M tokens, in / out
Run it yourself
Official weightsdeepseek-ai/DeepSeek-V3-0324↗Host prices from OpenRouter; going direct can differ.
Facts and sources
Released24 Mar 2025
Announcementnot linked yet
Release entry–
Scored asDeepSeek-V3 (Mar 2025)
Output speed–
Hugging Facedeepseek-ai/DeepSeek-V3-0324 ↗