Skip to main content

DeepSeek V4 Pro Hits SiliconFlow: 1M Context, Triple Reasoning Modes, and a Shockingly Low Cache Price

DeepSeek's newest model, V4 Pro, has officially arrived on SiliconFlow, and it's making waves with some impressive specs. The model, dubbed DeepSeek-V4-Pro-0813, launched on day zero, meaning it's available right away for developers and businesses to test and integrate.

So, what's new? Three key numbers stand out: a 1 million token context window, three levels of reasoning intensity (low, medium, and high), and a stronger emphasis on coding, tool calls, and agent workflows. The open-source license remains MIT, so you can still use it freely.

But the pricing is where things get interesting. For input tokens, you're looking at $1.32 per million, while output tokens cost $3.96 per million. However, the real bargain is cache-hit tokens, which are priced at just $0.44 per million. That's a steal, especially for developers who rely on repeated calls and long context inputs.

Why does this matter? Well, for agent workflows that involve lots of back-and-forth with the model, the cost of cache hits can quickly add up. With this low price, you can now handle long-context scenarios without worrying about your budget blowing up. It's a practical move that addresses a common pain point for developers.

The focus on coding and tool calls is also a smart play. As AI becomes more integrated into development pipelines, having a model that excels at these tasks is crucial. Whether you're building a chatbot, automating workflows, or just experimenting with AI, V4 Pro seems ready to handle the job.

But let's not get ahead of ourselves. While the specs are impressive, real-world performance will depend on how well it handles specific tasks. Early adopters will likely put it through its paces, and we'll see if it lives up to the hype.

For now, though, the launch is a clear signal that DeepSeek is serious about competing in the AI space. With a competitive price point and features tailored to developers, V4 Pro could be a game-changer for those looking to build sophisticated AI applications without breaking the bank.

If you're curious, you can head over to SiliconFlow to check it out. Whether you're a seasoned developer or just starting, this model might be worth a look.

Key Points

  • 1M context window: Handles massive amounts of text in one go.
  • Three reasoning levels: Low, medium, and high to balance speed and accuracy.
  • Cache-hit tokens at $0.44/M: A budget-friendly option for repeated calls.
  • MIT license: Open and free to use.
  • Focus on coding and agents: Tailored for modern AI workflows.