Moonshot apologises after Kimi K3.1 Pro roll-out stumbles as promotional scrutiny widens

Chinese AI start-up Moonshot AI apologized on Monday for the limitations imposed on users of its latest model, the K3.1 Pro, after a botched roll-out that has also raised concerns about its promotional practices and cybersecurity measures. In a post on its X account, Moonshot acknowledged the rate limits and unnecessary quota consumption on the K3.1 Pro subscription plan, offering to restore 25% of users' monthly quota as an adjustment.

This apology came three days after the company introduced the model, which users reported exhausting their five-hour usage allowance in just 10 minutes when attempting to build a basic 3D game demo. Moonshot, founded in 2023 and expected to reach US$1 billion in annual recurring revenue by the end of the year, unveiled the K3.1 Pro as its most powerful model yet.

However, user complaints about rate limits quickly flooded social media. The company also faced scrutiny over allegations that it had offered to pay online influencers to promote the model without disclosing the sponsorship, an issue that would violate social media platform policies and the US Federal Trade Commission's guidelines.

Moonshot did not address the allegations in its Monday apology. The company's open-weight models have helped China narrow the gap with US counterparts, with the K3.1 Pro weights set to be released by November 4. However, the open-weight version of the model will have its cyber-hacking capabilities downgraded due to the risk of misuse by malicious actors.