deepseek releases v4 pro and flash as mit-licensed open weights

what happened

• On April 24, 2026 DeepSeek announced 'DeepSeek-V4 Preview is officially live & open-sourced!', with weights published on Hugging Face.[1]
• The Hugging Face model cards for DeepSeek-V4-Pro and DeepSeek-V4-Flash each state the repository and model weights are licensed under the MIT License; V4-Pro has 1.6T total and 49B activated parameters, and V4-Flash 284B total and 13B activated.[2,4]
• Both models support a 1M-token context, which DeepSeek says is now the default across all its official services.[1,2]
• DeepSeek's API accepts the OpenAI ChatCompletions and Anthropic formats under the model names deepseek-v4-pro and deepseek-v4-flash.[1,3]
• DeepSeek said the legacy deepseek-chat and deepseek-reasoner names would be 'fully retired and inaccessible after Jul 24th, 2026, 15:59 (UTC Time).'[1]
• DeepSeek-V4-Pro reached general availability on the app, web and API on August 13, 2026, and on September 10 V4.1-Flash replaced V4 Flash under the model name deepseek-flash.[3]
• The Hugging Face page shows usage snippets for vLLM, SGLang and Docker Model Runner, and the model card recommends a context window of at least 384K tokens for the Think Max reasoning mode.[2]