Skip to content
AI Primer

DeepSeek

A family of open language models from DeepSeek

A family of large language and reasoning models developed by DeepSeek, including named generations and variants such as DeepSeek-V3 and DeepSeek-R1.

Pricing

Official site · Sep 8, 2026, 1:01 PM
Input / 1M
CN¥3.00
Output / 1M
CN¥9.00
Cached input / 1M
CN¥0.10

Peak period is Beijing time Monday–Friday 09:00–12:00 and 14:00–18:00; off-peak rates are CNY 1.50 input (cache miss), CNY 0.05 input (cache hit), and CNY 4.50 output per 1M tokens. Other DeepSeek-V4 API models have separate published rates.

DeepSeek’s official API pricing lists token billing for the current DeepSeek-V4 lineup. This finding uses the explicitly stated peak-period rates for the text-capable deepseek-v4-flash API model: CNY 3.00 per 1M cache-miss input tokens, CNY 0.10 per 1M cache-hit input tokens, and CNY 9.00 per 1M output tokens. The official table also states off-peak rates are half these rates.

View source

Model Intelligence

Arena ranking
7
Benchmarkable
No
Model level
family
Intelligence Index
4.7
MATH-500
0.74

Recent stories

1 linked story
AI PrimerAI Primer

Your daily guide to AI tools, workflows, and creative inspiration.

© 2026 AI Primer. All rights reserved.