Industry · Status at source date: Newsletter archive
Moonshot releases Kimi K3 as the largest open-weight model yet
Moonshot AI has released Kimi K3, a 2.8-trillion-parameter open-weight model with a one-million-token context window and native visual understanding. Moonshot says it trails Claude Fable 5 and GPT-5.6 Sol on overall performance but consistently beats Claude Opus 4.8 and GPT-5.5 on coding, visual understanding, and long-horizon tasks, and it led Arena's front-end programming leaderboard within a day of launch.
What changed
Full model weights are due by 27 July.
Price is the part that matters commercially: K3 runs at $15 per million output tokens, compared to $50 for Fable. Bank of America noted that Moonshot reached this level despite limited access to advanced chips, and shares in Chinese rivals Zhipu and MiniMax fell 28.4% and 15.6% on the news.
Why it matters for advertisers
For you, this means the cost floor for capable models is dropping fast, so if you run AI over large volumes of ad copy, feeds, or reporting, retest what an open-weight model can now do at a fraction of frontier pricing.
Perspective from the original PMC newsletter.Sources & contributor credit
- Newsletter coverage · Paid Media Collective newsletter
Original newsletter text, contributor labels and media for this update.
- Source referenced in newsletterVentureBeat
Linked from the original newsletter. The source publication date has not been independently confirmed.
- Source referenced in newsletterCNBC
Linked from the original newsletter. The source publication date has not been independently confirmed.
Newsletter curated by Yoann Ferrand. The earliest individual contributor has not been verified for this update.
Original creator unverified
The original creator has not yet been verified. Newsletter curation and publication do not establish original authorship.
Attribution evidence and limitations
Full attribution review pending.
A source link alone does not establish original authorship.
Original image and video creators have not yet been verified.
- Published on this site
This update reflects the dated source reporting. Availability may have changed. Further coverage of this same development will be added to this page.
Original newsletter text and archive evidence
Moonshot releases Kimi K3 as the largest open-weight model yet
Moonshot AI has released Kimi K3, a 2.8-trillion-parameter open-weight model with a one-million-token context window and native visual understanding. Moonshot says it trails Claude Fable 5 and GPT-5.6 Sol on overall performance but consistently beats Claude Opus 4.8 and GPT-5.5 on coding, visual understanding, and long-horizon tasks, and it led Arena's front-end programming leaderboard within a day of launch. Full model weights are due by 27 July.
Price is the part that matters commercially: K3 runs at $15 per million output tokens, compared to $50 for Fable. Bank of America noted that Moonshot reached this level despite limited access to advanced chips, and shares in Chinese rivals Zhipu and MiniMax fell 28.4% and 15.6% on the news. For you, this means the cost floor for capable models is dropping fast, so if you run AI over large volumes of ad copy, feeds, or reporting, retest what an open-weight model can now do at a fraction of frontier pricing.
Source captured . No explicit first-contributor label was provided for this update.
