DeepSeek Cuts Flagship AI Model Prices by 75% as Funding Round Looms
Listen to the full version
Chinese artificial intelligence startup DeepSeek has permanently cut the application programming interface (API) price of its flagship DeepSeek-V4-Pro model by 75%, escalating competition with both domestic and overseas rivals.
The company said on May 22 that a temporary promotional discount for developers would become its standard pricing tier. The move lowers the cache-miss input cost for the V4-Pro model to 3 yuan ($0.44) per million tokens, less than one-tenth of the roughly $5 charged for OpenAI’s GPT-5.5 and well below the $0.95 charged by Chinese rival Kimi.
Unlock exclusive discounts with a Caixin group subscription — ideal for teams and organizations.
Save an extra $50. Introductory offer for new readers. Subscribe now.
- RELATED
- DIGEST HUB
Digest Hub
Back
- DeepSeek permanently cut V4-Pro API price by 75% to 3 yuan/million tokens, undercutting OpenAI’s GPT-5.5 and rival Kimi.
- Company seeks first external funding at $44B valuation; market share on OpenRouter rose to 23.1%.
- V4-Pro is largest open-weight model (1.6T parameters) but ranks 9th globally with 63.87% accuracy.
- ...
- {"chinese":"中国人工智能初创公司DeepSeek将其旗舰模型DeepSeek-V4-Pro的API价格永久下调75%，降至每百万token 0.44美元，远低于OpenAI GPT-5.5的5美元和Kimi的0.95美元。该公司正以3000亿元人民币估值寻求首轮外部融资。","english":"Chinese AI startup DeepSeek permanently cut the API price of its flagship V4-Pro model by 75% to $0.44 per million tokens, far below OpenAI's GPT-5.5 at $5 and rival Kimi at $0.95. The company is seeking its first external funding round at a $44 billion valuation."}
- late April 2026：
- DeepSeek releases DeepSeek-V4-Pro, the world's largest open-weight AI model with 1.6 trillion parameters.
- May 22, 2026：
- DeepSeek permanently cuts the API price of DeepSeek-V4-Pro by 75%, making a temporary promotional discount the standard pricing.
- May 2026 (last week)：
- DeepSeek-V4-Flash model ranks first on OpenRouter with 3.43 trillion tokens consumed.
- May 2026：
- Third-party benchmark provider VALS AI ranks DeepSeek-V4-Pro ninth globally with an average accuracy score of 63.87%.
- MOST POPULAR
Sign in with Google
Sign in with Facebook
Sign in with 财新