OPEN LINK ↗
// 1d agoMODEL RELEASE
DeepSeek V4 Pro Adds Peak Pricing
DeepSeek has released V4-Pro generally with stronger agent performance, adjustable reasoning effort, and native OpenAI Responses API support. New API pricing begins August 16, with peak-hour rates doubling off-peak prices.
// ANALYSIS
DeepSeek is trading some of its low-cost appeal for infrastructure-aware pricing, making scheduling a real optimization lever for developers.
- –V4-Pro targets production agent workflows with improved tool use and coding performance
- –Native Responses API support lowers integration friction for Codex and OpenAI-compatible stacks
- –Off-peak pricing is 50% cheaper, but peak output costs reach $3.96 per million tokens for V4-Pro
- –Peak hours run 01:00–04:00 and 06:00–10:00 UTC, rewarding batch workloads and flexible scheduling
- –Developers should update cost models, monitor cache-hit rates, and avoid assuming DeepSeek remains uniformly cheap
// TAGS
deepseek-v4-prollmapiinferencepricingagenttool-usecoding-agent
DISCOVERED
1d ago
2026-08-14
PUBLISHED
1d ago
2026-08-14
RELEVANCE
9/ 10
AUTHOR
fagnerbrack