Announcement: DeepSeek-V4-Pro General Availability & Pricing Shift

Breaking News: We are thrilled to announce that DeepSeek-V4-Pro is officially available for general use!

🚀 Introducing DeepSeek-V4-Pro

The new V4 lineup introduces a more dynamic approach to computation. Users can now adjust the reasoning effort based on the specific needs of their task:

• Low: Ideal for straightforward, simple queries.
• High: Tailored for standard daily Agent-based workflows.
• Max: Reserved for the most intricate and complex problem-solving.

Reasoning Effort Logic

Integration & Compatibility

We have implemented native support for the OpenAI Responses API, which has been specifically tuned for Codex.

Setup Checklist:

• Review API documentation
• Perform one-click setup
• Verify model_name (Note: names remain unchanged)

To integrate, you can use a configuration similar to this:

{
  "model": "deepseek-v4-pro",
  "reasoning_effort": "high",
  "api_version": "2026-08-16"
}

---

💰 API Pricing Overhaul

We are transitioning from flat-rate pricing to a tiered system based on demand. This allows users to optimize their costs by scheduling heavy workloads during quieter periods.

Rate Structure

The cost difference is defined by the following relationship: $\text{Off-Peak Price} = \text{Peak Price} \times 0.5$

Effective Date: The new pricing schedule commences at 16:00 UTC on August 16, 2026.