DeepSeek’s new math mannequin stirs buzz about its mysterious next-gen LLM, R2 – Firstpost

Chinese language AI startup DeepSeek has dropped a shock improve to its math-focused language mannequin, intensifying hypothesis round its upcoming next-generation reasoning system recognized merely as R2.

Whereas the corporate has remained tight-lipped concerning the new mannequin, the sudden launch of Prover-V2, a 671-billion-parameter mannequin fine-tuned for mathematical proof-solving, has reignited on-line chatter throughout developer and investor communities alike.

The brand new mannequin, primarily based on DeepSeek’s V3 basis, was quietly open-sourced on Wednesday (April 30). It builds on Prover-V1.5, which launched final August and drew curiosity from academia and aggressive math circles.

Whereas Prover-V2 is just not the long-awaited R2, it has been broadly interpreted as a key stepping stone. Customers on X and Reddit are calling it a math capacity improve laying the groundwork for what may very well be the following leap in reasoning-focused LLMs from China’s most-watched AI startup,
South China Morning Put up reported.

Based in 2023 by Liang Wenfeng as a spinout of his quantitative hedge fund Excessive-Flyer, DeepSeek shortly gained world consideration with its R1 mannequin, launched in January. R1 surprised the AI world by matching OpenAI’s o1-level efficiency at a fraction of the fee, all whereas utilizing far fewer assets. That success set expectations sky-high for no matter comes subsequent.

No timeline for R2

Nevertheless, DeepSeek has provided no public timeline for R2. The corporate has revealed little past analysis papers and mannequin updates, fuelling a vacuum of knowledge that has been crammed by social media hypothesis. One viral submit from a DeepSeek researcher merely saying Prover-V2 led to a cascade of replies pleading for an R2 launch. “R2 R2 R2 please,” one person wrote.

Much more buzz got here from Chinese language stock-trading boards like Jiuyangongshe, the place rumors of an imminent R2 drop spilled over into Western platforms. A notable US enterprise capital investor picked up the chatter on X, propelling the information into wider investor circles. Searches for “DeepSeek” and “R2” have spiked on Google Traits over the previous week.

Including to the intrigue, DeepSeek is now quietly ramping up hiring. The corporate lately posted openings for its first product and design lead, primarily based in both Beijing or Hangzhou. The job description requires constructing a “next-generation clever product expertise” rooted in LLM tech. The startup can be actively recruiting a chief monetary officer and chief working officer.

Competitors in China rising

This comes simply as different main Chinese language corporations are upping their recreation. On Tuesday, Alibaba unveiled Qwen3, its newest household of fashions that the corporate says outperform DeepSeek-R1 on a number of metrics. The announcement was seen by some as a shot throughout the bow, upping the stress on DeepSeek to ship a follow-up.

In the meantime, in america, OpenAI lately launched o3 and o4-mini, touting them as its “most succesful fashions thus far.” Whereas DeepSeek lacks entry to cutting-edge Nvidia chips as a result of US export restrictions, it has constructed a repute for maximising efficiency on constrained {hardware}, drawing curiosity from technologists and policymakers alike.

The launch of Prover-V2 might not be the generational leap that some had been hoping for, however it suggests DeepSeek is much from idle. With the corporate scaling up and hype constructing quick, the query now is just not whether or not R2 is coming, however how shut we’re to seeing it in motion.