Chinese AI startup Moonshot AI has released Kimi K3, its largest and most capable open-source model to date, with a particular focus on coding performance. The release positions Kimi K3 among the most parameter-dense publicly available models globally, cutting into a capability gap that US labs have held for the past two years. Sources differ on whether it is strictly a coding model or a general-purpose frontier model, but agree it meaningfully closes the distance between Chinese and American AI.
Every credible Chinese open-source release puts fresh pressure on the premium valuations US AI companies carry. If frontier-level AI becomes freely available from overseas, the moat for closed US models narrows, and investors in US AI software names may need to reassess how durable those moats really are. Hardware suppliers are less directly affected, since training large models still requires chips.
Next major US AI lab model releases (OpenAI GPT-5 expected mid-2025). Moonshot AI benchmark scores on standard coding and reasoning tests, published alongside or shortly after the release. US export control review meetings, next scheduled review window around Q3 2025.
Full analysis · Subscribers
The deep dive (bull case, bear case, and the data point that decides which side wins), the cause-and-effect chain behind the move, plain-English explainers for every block, and the live update timeline (2 updates so far).
Aggregated reads dozens of sources in five languages and turns the day into plain-English cards like this one.
Educational analysis of public information, not investment advice. Report an error · Corrections policy
← Today's brief