DeepSeek slashes V4-Pro price

DeepSeek cut the API price of its new DeepSeek-V4-Pro model by 75% on April 27, with the discount set to run until May 5. (api-docs.deepseek.com) (money.usnews.com) The company’s pricing page lists V4-Pro at $0.435 per million input tokens, $0.87 per million output tokens, and $0.003625 per million cached input tokens during the promotion. The regular listed prices are $1.74, $3.48, and $0.0145. (api-docs.deepseek.com) DeepSeek also cut input cache-hit prices across all of its models to one-tenth of their launch levels. Cache hits are repeated prompts the system can reuse instead of processing from scratch. (api-docs.deepseek.com) (bloomberg.com) V4-Pro only entered the DeepSeek API on April 24, alongside V4-Flash. DeepSeek said the new models are available through both OpenAI-style and Anthropic-style interfaces, while older names like `deepseek-chat` and `deepseek-reasoner` are set to be retired on July 24. (api-docs.deepseek.com 1) (api-docs.deepseek.com 2) That means the price cut landed just days after launch, before most developers would have finished testing the model in production. Bloomberg said the move adds pressure in a Chinese artificial intelligence market already competing hard on price. (bloomberg.com) (money.usnews.com) DeepSeek’s docs describe V4-Pro and V4-Flash as models with a one-million-token context window and a maximum output of 384,000 tokens. In plain terms, that is a very large text buffer for long documents, codebases, and multi-step agent workflows. (api-docs.deepseek.com) Reuters reported that DeepSeek announced the cuts in a post on X. The company’s own pricing page says the promotional V4-Pro rates expire at 15:59 Coordinated Universal Time on May 5, 2026. (money.usnews.com) (api-docs.deepseek.com) The immediate result is simple: a model released on April 24 got cheaper again on April 27, and DeepSeek used both headline discounts and lower cache pricing to push adoption fast. (api-docs.deepseek.com 1) (api-docs.deepseek.com 2)