Finance

China's DeepSeek slashes prices for new AI model

Published by Global Banking & Finance Review

Posted on April 27, 2026

1 min read

· Last updated: April 27, 2026

Add as preferred source on Google
China's DeepSeek slashes prices for new AI model
Global Banking & Finance Awards 2026 — Call for Entries

April 27 (Reuters) - China's DeepSeek is offering developers a 75% discount on its newly unveiled AI model, DeepSeek-V4-Pro, until May 5. The company is also cutting prices for input cache hits across

DeepSeek Launches AI Model V4 Pro With Major Price Cuts and API Savings

DeepSeek-V4-Pro Launch Details and Pricing Strategy

Launch Announcement and Discount Offer

April 27 (Reuters) - China's DeepSeek is offering developers a 75% discount on its newly unveiled AI model, DeepSeek-V4-Pro, until May 5.

API Price Reductions

The company is also cutting prices for input cache hits across its entire DeepSeek API lineup to one-tenth of the original price, it said in a post on X.

Model Preview and Technical Adaptation

V4 Model Preview

On Friday, DeepSeek launched a preview of its highly anticipated V4 model, which has been adapted for Huawei's chip technology.

Model Variants

V4 comes in two versions: the more powerful and higher priced Pro, and the lighter, cheaper Flash variant.

Performance and Use Cases

Benchmark Performance

The Pro version outperforms other open-source models in world-knowledge benchmarks, trailing only Google's closed-source Gemini-Pro-3.1, DeepSeek said.

AI Agent Suitability

Complex Task Execution

According to the Chinese startup, the V4 models are particularly suited to AI agent work, which can execute more complex tasks than chatbots but require greater computing power.

(Reporting by Disha Mishra in Bengaluru; Editing by Sumana Nandy)

Key Takeaways

  • DeepSeek‑V4‑Pro, a 1.6 trillion‑parameter Mixture‑of‑Experts model with 1M‑token context, outperforms all open models on benchmarks and rivals top closed‑source ones while being fully open‑source under an MIT license (apnews.com).
  • The API cache hit pricing has been reduced to one‑tenth of the original cost, enabling massive cost savings in usage thanks to context reuse – a promotion confirmed in a developer post (reddit.com).
  • This pricing strategy underscores DeepSeek’s push into AI agent workloads, with the V4‑Pro tailored for complex task execution and optimized for Huawei Ascend chips, reducing reliance on U.S. hardware and positioning it as a competitive, cost‑effective model in the global AI race (tomshardware.com)

References

Frequently Asked Questions

What discount is DeepSeek offering on its new AI model?
DeepSeek is offering developers a 75% discount on its DeepSeek-V4-Pro AI model until May 5.
What are the versions of DeepSeek's new AI model?
DeepSeek's new V4 model is available in two versions: the powerful Pro and the lighter, cheaper Flash.
How does DeepSeek-V4-Pro perform compared to other AI models?
DeepSeek-V4-Pro outperforms other open-source models in world-knowledge benchmarks, trailing only Google's Gemini-Pro-3.1.
Has DeepSeek changed pricing for any other products?
Yes, DeepSeek has cut prices for input cache hits across its API lineup to one-tenth of the original price.
What is special about DeepSeek's V4 models?
The V4 models are adapted for Huawei chip technology and are particularly suitable for AI agent work requiring more computing power.

Tags

Related Articles

More from Finance

Explore more articles in the Finance category