Artificial intelligence startup DeepSeek has unveiled its latest open-source models, DeepSeek-V4-Pro and DeepSeek-V4-Flash, on April 24, 2026. The models feature an unprecedented 1-million token context window, enabling them to process and understand extremely lengthy documents with greater nuance.
DeepSeek-V4-Pro is setting new standards in open-weight AI, leading all competitors in agentic coding benchmarks. It demonstrates superior performance in automated code generation and achieved a score of 1554 on GDPval-AA, a metric for real-world work task proficiency.
By releasing these models openly, DeepSeek is democratizing advanced agentic capabilities, allowing developers worldwide to access cutting-edge AI tools. This move directly challenges proprietary models by offering powerful, accessible alternatives.