i
News
News · 2026-11-16

Moonshot AI shifts Kimi from coding to scientific research

Dataist Dataist

Moonshot AI is shifting Kimi from coding to science: benchmark progress is slowing

Cover: Moonshot AI shifts Kimi from coding to scientific research

Moonshot AI is steering its Kimi language models toward research in biology, chemistry and pharmaceuticals, saying there is less room for further progress in coding. At Web Summit in Lisbon, Moonshot AI’s international director, Giacomo Ficari, said science needs smarter models than law, finance or programming. Meanwhile, Beijing-based DP Technology is incorporating Kimi K3.1 Pro into its BioMaster research assistant. In one test, the model analyzed gene data from 3,696 mouse pancreatic cells and identified 100 candidate genes in 90 minutes — work estimated to take a human four hours.

Moonshot AI is shifting Kimi from coding to scientific research

Results and limitations

The ability to make scientific discoveries independently remains a distant prospect. On TCSExam, a benchmark of 100 theoretical computer science problems, leading models were able to produce verifiable proofs but did not independently solve a single open conjecture. Similar limitations have been noted among US competitors.

In September, Anthropic’s Claude discovered an enzyme system. Mathematicians, meanwhile, criticized OpenAI’s claims of breakthroughs in mathematics and physics.

The pharmaceutical industry is also unsure of AI’s practical benefits. A June Citi survey of 50 US research executives found that 72% already use AI widely, but only 24% expect it to significantly improve the success rate of drug development.

Dataist’s take: Kimi’s move into scientific research shows where Moonshot AI is looking for further growth as progress in coding slows. But the BioMaster test results so far point to help with data analysis, not independent discoveries; the Citi survey also suggests most respondents do not expect AI to significantly affect the success rate of drug development.

Daily AI news

Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.

Only what matters — every day

Follow on X