DeepSeek V4 official launch mid-July with peak-hour pricing
Preview users get 2x rates during Beijing business hours starting mid-July.
DeepSeek's official announcement on July 1, 2026, confirms that V4 will officially launch in mid-July 2026. This is not a new model name but the graduation from the preview phase, bringing feature optimizations and performance improvements. The key change is the introduction of peak-hour pricing: 2× the baseline rate during defined Beijing business windows, while off-peak rates will stay the same as today's preview pricing. Users who have been calling their models deepseek-v4-pro or deepseek-v4-flash since April 2026 were actually using the preview, not the official release.
DeepSeek expects heavy demand at the official launch, justifying the new pricing structure. Teortaxes (@teortaxesTex) bluntly noted that while users thought they had V4 for over two months, it was only the preview. The move mirrors industry trends where AI providers dynamically price API access during high-demand periods. For developers and businesses relying on DeepSeek, this means planning workloads around off-peak hours to avoid cost spikes. The official release documentation and migration guide are already live.
- DeepSeek V4 official launch is mid-July 2026, not a new model but graduation from preview.
- Peak-hour pricing introduced: 2× baseline during Beijing business windows (8am-6pm Beijing time).
- Off-peak rates remain unchanged from current preview pricing; preview versions have been available since April 2026.
Why It Matters
Dynamic pricing for AI APIs makes cost planning critical; developers must shift heavy usage to off-peak hours.