Models

DeepSeek V3 (Dec '24)

DeepSeek V3 is a powerful and cost-effective large language model released in December 2024. It excels in complex reasoning, coding, and long-context understanding (up to 128K tokens), while maintaining high inference speed and low operational costs.

CodingReasoningFastCheapLong context
Input / 1M tokens
$0.40
Output / 1M tokens
$0.89
Supported plans
6

Benchmark history

Evaluations

15

TAU2

Measured May 14, 2026Source

Score

0.23

Terminalbench Hard

Measured May 14, 2026Source

Score

0.07

Lcr

Measured May 14, 2026Source

Score

0.29

Ifbench

Measured May 14, 2026Source

Score

0.35

Aime 25

Measured May 14, 2026Source

Score

0.26

Aime

Measured May 14, 2026Source

Score

0.25

Math 500

Measured May 14, 2026Source

Score

0.89

Scicode

Measured May 14, 2026Source

Score

0.35

Livecodebench

Measured May 14, 2026Source

Score

0.36

Hle

Measured May 14, 2026Source

Score

0.04

Gpqa

Measured May 14, 2026Source

Score

0.56

Mmlu Pro

Measured May 14, 2026Source

Score

0.75

Artificial Analysis Math Index

Measured May 14, 2026Source

Score

26

Artificial Analysis Coding Index

Measured May 14, 2026Source

Score

16.4

Artificial Analysis Intelligence Index

Measured May 14, 2026Source

Score

16.5

Plan availability

Products and plans that support this model

3

Discussion

Thinking... Make sure you are connected to GitHub server