# DeepSeek unveiled the V4 Flash and V4 Pro models along with fairly basic details on pricing, parameters and the ability to process a maximum output of 384,000 tokens.

*semiconductor · news · 2026-04-24 · The Straits Times*

## Key points

- DeepSeek has launched its V4 Flash and V4 Pro AI models with a 384,000 token output limit.
- The R1 model by DeepSeek triggered a 2025 trillion-dollar stock market sell-off.
- R1 matched top AI system performance while being built for a fraction of the usual cost.
- US tech giants are projected to spend US$650 billion on AI infrastructure in 2026.

**Companies:** DeepSeek
**Countries:** China, United States

[Read the full story on The Straits Times](https://www.straitstimes.com/business/companies-markets/chinas-deepseek-unveils-new-flagship-ai-model-a-year-after-breakthrough)

---

Canonical: https://newsio.io/n/27e720ff-d330-465b-b290-e7110596d1a9/deepseek-unveiled-the-v4-flash-and-v4-pro-models-along-with-fairly-basic-details
Summarized by Newsio from The Straits Times. https://newsio.io/how-it-works