DeepSeek announced that that its DeepSeek-V4-Pro API is "75% OFF" through May 5, 2026 at 15:59 UTC, while also listing several integration updates aimed at agent and coding workflows.
In the post, the company mentions that Claude Code users can set the model to `deepseek-v4-pro[1m]` to unlock "1M context." It also points to updates for OpenCode, which DeepSeek says should be on v1.14.24 or later, and OpenClaw, which it says should be updated to v2026.4.24 or later. DeepSeek directed users to its official API docs for the full details.
The announcement drew a mix of reactions from developers and other users. Some called the model a strong value proposition, while others focused on pricing math, long-context usage, and whether DeepSeek can remain profitable on the lower rates. A few commenters also asked practical setup questions about Claude Code, Visual Studio, and whether the discount applies to usage or only to account top-ups.
Several replies centered on the long-context angle, with users pointing to the 1M-token window as useful for code review, troubleshooting, and large engineering reviews. Others highlighted cache-hit pricing and suggested that the discounted rate could change how teams route model usage across assistant workflows.
DeepSeek’s post also arrived alongside one recurring theme in the replies: the idea that cheaper model access is increasingly being treated as a configuration choice inside agent stacks rather than a fixed default. One commenter put it bluntly, calling the packaging for Claude Code and OpenClaw-style workflows the part worth paying attention to.
Source: DeepSeek on X
