Moonshot’s upcoming Kimi 3 is expected to close the gap with Anthropic’s Opus 4.8
TechCrunch Dominic-Madori Davis ● Covered by 75 sources
Moonshot AI's next Kimi model reportedly rivals Anthropic's Opus 4.8, sources tell the FT. A Chinese open-weight model closing in on the priciest US labs — that's a big shift in leverage.
Based on reporting by TechCrunch, Dominic-Madori Davis — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Moonshot AI isn't messing around with its next release. According to the Financial Times, sources familiar with the matter say Kimi K3 will perform on par with, or possibly beat, Anthropic's Opus 4.8 — currently one of the most capable closed models on the market. That's a striking claim for a lab that, until recently, was mostly known in open-source circles rather than boardrooms.
The numbers involved are enormous. K3 is expected to land somewhere between 2 trillion and 3 trillion parameters, which would make it the largest open-weight model to come out of China. Moonshot's prior release, Kimi K2, already turned heads by ranking near the top of several benchmarks while staying open, and reportedly K3 is built to push that gap even tighter against OpenAI and Anthropic's frontier systems. FT says the release could land within days, not months.
Money is following the momentum. Moonshot is said to be closing a new funding round that would value the company at $31.5 billion — a sharp jump from the $20 billion valuation it hit in May after raising $2 billion. Investors clearly think an open model that can trade blows with Opus-tier systems is worth betting big on, even in a market crowded with well-funded rivals.
The timing lines up with a broader argument brewing across the industry: is it still worth paying premium prices for closed AI products like ChatGPT or Claude? Some executives worry labs could extract value from the very data customers feed into these tools, and they're increasingly pointing customers toward cheaper, self-hosted alternatives — DeepSeek, Z.ai, and now Moonshot among them. If Kimi K3 really closes the gap with Opus 4.8, that argument gets a lot harder to dismiss.
My take — AI-written commentary, not fact-checked reporting
I run an independent AI news site, so I have zero patience for the idea that frontier intelligence has to live behind a subscription wall — and Kimi K3, if it delivers, is exactly the kind of pressure that keeps OpenAI and Anthropic honest on pricing. Closed labs love to frame open models as perpetually 'close but not quite there'; every few months that gap gets smaller, and eventually somebody's going to have to admit it closed. Bet on China's open-weight labs, not because they're altruistic, but because competition is the only thing that's ever actually lowered AI prices.
Read more about this at: TechCrunch