DeepSeek Accelerates AI Model Timeline as Market Reacts To Low-Cost Breakthrough - Slashdot
Feb 25, 2025 - slashdot.org
Chinese AI startup DeepSeek is accelerating the release of its R2 model after the success of its R1 model, which outperformed many U.S. competitors at a significantly lower cost, causing a market selloff exceeding $1 trillion. Originally planned for a May release, the Hangzhou-based company now aims to launch R2 "as early as possible," according to a Reuters report. The new model is expected to offer enhanced coding capabilities and multilingual reasoning. DeepSeek's competitive edge is largely due to its parent company High-Flyer's early investment in computing power, including two supercomputing clusters with Nvidia chips acquired before U.S. export restrictions.
DeepSeek's cost-efficiency is attributed to innovative architecture choices such as Mixture-of-Experts (MoE) and multihead latent attention (MLA). Analysts from Bernstein note that DeepSeek's pricing is 20-40 times cheaper than comparable models from OpenAI. This competitive pressure has led OpenAI to reduce prices and release a scaled-down model, while Google's Gemini has introduced discounted access tiers.
Key takeaways:
DeepSeek is accelerating the release of its R2 model after the success of its R1 model, which outperformed US competitors at a lower cost.
The R2 model will offer improved coding capabilities and reasoning in multiple languages beyond English.
DeepSeek's competitive advantage is due to early investments in computing power, including supercomputing clusters with Nvidia A100 chips.
DeepSeek's models are 20-40 times cheaper than OpenAI's, leading to competitive pressure on OpenAI and Google to adjust their pricing and offerings.
