DeepSeek opened preview access to DeepSeek-V4-Pro (1.6T total parameters, 49B activated) and DeepSeek-V4-Flash (284B total, 13B activated) on April 24, 2026, both supporting a 1M-token context window. At the 1M-token context length, V4-Pro requires only 27% of the single-token inference FLOPs and 10% of the KV cache of DeepSeek-V3.2. DeepSeek says V4-Pro reaches top-tier coding benchmark performance and narrows the gap with leading closed-source models on reasoning and agentic tasks. Pricing has not been announced for the preview period.