DeepSeek V4 Pro 0813 Is Here — Who’s Going to Try It?

DeepSeek has updated V4-Pro to the new 0813 version.

V4-Pro is the larger model in the V4 family, designed for more demanding reasoning, coding, and agent workloads, while still keeping the 1M context window.

What I’m most interested in is how it performs in real development compared with models like GPT, Claude, and Gemini — especially when cost is taken into account.

We’ve already added the latest DeepSeek V4 Pro to ApiHub.

If you want to test it without setting up another separate provider account, new ApiHub users get free credits, so you can try the model first and compare it with other models through the same API.

Curious to hear from developers here:

What would you test first with DeepSeek V4 Pro — coding, agents, reasoning, or something else?