DeepSeek-V4-Pro-0813 (GA)

DeepSeek-V4-Pro-0813 (GA) is the general availability release of deepseek-v4-pro on app, web, and API, featuring selectable reasoning effort (low/high/max, also on V4-Flash), native OpenAI Responses API support optimized for Codex, and an Expert Mode in the app/web. Note that flat pricing ends on August 16, 2026 at 16:00 UTC, replaced by peak/off-peak tiering (peak 01:00–04:00 and 06:00–10:00 UTC; off-peak 50% cheaper), with V4-Pro peak rates at $0.044 per 1M cache-hit input tokens, $1.32 per 1M cache-miss input tokens, and $3.96 per 1M output tokens.