TLDR
- Moonshot AI, a Chinese startup, released Kimi K3 as a free open-weight AI model with performance comparable to leading American systems from OpenAI and Anthropic
- Within 48 hours of release, overwhelming demand forced Moonshot to halt new user registrations
- White House technology advisor Michael Kratsios claimed Moonshot used distillation techniques to replicate Anthropic’s Fable model illegally
- Both Moonshot AI and Chinese diplomatic representatives rejected the accusations as baseless
- Treasury Secretary Scott Bessent indicated potential sanctions and trade restrictions could target the company
On Friday, Chinese artificial intelligence company Moonshot AI unveiled Kimi K3, branding it as an open-weight model with 2.8 trillion parameters. According to the firm, it achieves performance levels comparable to Fable from Anthropic, representing one of America’s most sophisticated AI platforms.
🚨JUST IN: White House technology chief Michael Kratsios accuses Chinese AI startup Moonshot of illegally accessing US technology to develop its Kimi K3 model.
Kratsios alleges the company obtained servers using NVIDIA GB300 chips and routed workloads through infrastructure in… https://t.co/i2M8rPDdY0 pic.twitter.com/GBZhvuRcVr
— Coin Bureau (@coinbureau) July 22, 2026
In contrast to most American models requiring paid subscriptions, Kimi K3 is available for free download, modification, and self-hosting. This accessibility immediately appealed to software developers across the globe.
The overwhelming response forced Moonshot to pause new registrations just 48 hours after launch. The infrastructure simply couldn’t handle the traffic surge.
Benchmark testing showed Kimi K3 achieving comparable results to OpenAI’s GPT series and Anthropic’s Claude models. Certain evaluations showed it exceeding their capabilities.
Research firm Counterpoint estimates the performance difference between Chinese and American AI systems has shrunk to approximately three to six months.
Washington Alleges Intellectual Property Violations
The model’s debut prompted immediate criticism from US officials. Michael Kratsios, a White House technology advisor, claimed via X that Moonshot used distillation methods on Anthropic’s Fable model to create K3, characterizing it as “large-scale, covert industrial distillation.”
Sarah Heck, Anthropic’s policy director, supported these allegations, labeling the situation as “IP theft and industrial espionage” with “serious national security risks.”
Scott Bessent, serving as Treasury Secretary, mentioned that sanctions and inclusion on trade restriction lists are under consideration for companies engaging in intellectual property violations.
Moonshot firmly rejected these charges. China’s embassy spokesperson in Washington dismissed the allegations as “entirely unfounded.”
Many independent experts expressed doubt about the distillation claims. Wei Sun from Counterpoint questioned the feasibility of the timeline. According to Sun, Moonshot wouldn’t have had sufficient time to develop K3 using Anthropic’s recent models before the product launch.
Company Origins and Future Implications
Yang Zhilin established Moonshot three years ago. The Chinese technologist, educated at Carnegie Mellon University, famously declined employment opportunities linked to Apple’s Tim Cook to pursue his entrepreneurial vision.
Throughout 2023, the startup secured approximately $300 million across two major financing rounds.
The Kimi K3 release mirrors the trajectory established by DeepSeek earlier in 2025, when that Chinese model disrupted industry expectations. Experts suggest China’s AI advancement is demonstrating greater consistency and credibility.
“DeepSeek was a surprise. K3 is evidence that China’s progress is becoming more repeatable,” said Poe Zhao, founder of Hello China Tech.
The timing is notable, arriving just one month following the US government’s withdrawal of Anthropic’s Fable and Mythos models citing security considerations. China’s transparent AI development environment continues accelerating rapidly.





