Skip to content

DeepSeek V3

ModelOpen sourceActive
by DeepSeek · family “deepseek-v”

DeepSeek-V3 is the latest model from the DeepSeek team, building upon the instruction following and coding abilities of the previous versions. Pre-trained on nearly 15 trillion tokens, the reported evaluations... (description from the OpenRouter listing)

in: textout: textTool useStructured output
Epoch Capabilities Index
132.4 #145 of 268
90% CI 126.4 – 135.6
Listing ↗
Released
26 Dec 2024
Context window
164K tokens
Max output
16K tokens
Input $ / 1M tokens
$0.26
Output $ / 1M tokens
$1.03
Cached input $ / 1M
Unknown
Each benchmark shown separately with its own source

Benchmark results (14)

BenchmarkDomainScorevs best recordedSettingRunSource
METR Time Horizonsagents47.4%
56%
——External ↗
FrontierMath-2025-02-28-Privatesupersededmath1.7% ±0.8
3%
—7 Mar 2025Epoch ↗
Aider polyglotcoding48.4%
55%
——External ↗
OTIS Mock AIME 2024-2025math15.8% ±4.3
16%
—25 Feb 2025Eval log ↗
SimpleBenchreasoning18.9%
23%
——External ↗
GPQA diamondscience56.5% ±2.8
59%
—27 Jan 2025Eval log ↗
BBHreasoning87.5%
98%
——External ↗
MATH level 5math64.9% ±1.0
66%
—27 Jan 2025Epoch ↗
MMLUknowledge87.2%
99%
——External ↗
PIQAother84.7%
95%
——External ↗
Winograndeother85.2%
96%
——External ↗
HellaSwagreasoning88.9%
93%
——External ↗
ARC AI2other95.3%
100%
——External ↗
TriviaQAother82.9%
95%
——External ↗

Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.

API list price over time

Price history

$ per 1M tokens

13 recorded prices on 1 Oct 2025 (OpenRouter listing + Internet Archive snapshots); re-read on every data refresh, most recently 11h ago. Steps show when the price changed.

Events

No events recorded yet.