Editorial illustration for Moonshot AI's K3 Models Generate 300 Billion Daily Tokens
Moonshot AI's K3 Models Hit 300B Daily Tokens
Moonshot AI wants to double its money in four months. Bloomberg reported Friday that the Beijing lab, best known for its Kimi chatbot and open-weight K3 models, is targeting $2 billion in annualized revenue by year's end, up from its reported run rate in August. The number tracks with what's happened on OpenRouter since K3 launched this summer: the platform currently shows as many as 300 billion tokens a day moving through K3 models, even after usage cooled somewhat from its peak.
That's still small next to OpenAI's roughly $40 billion revenue figure or Anthropic's $65 billion, and Moonshot's margins are thinner by design since anyone can download and run K3's weights for free. But the projection suggests open-weight models can generate real revenue, not just downloads and goodwill.
The growth comes with a catch. Moonshot's approach to building K3 has drawn scrutiny this week, with Anthropic leveling specific accusations about how the model actually got made, and what data trained it.
One of China’s most prominent AI labs, Moonshot AI, believes it can turn its popular open-weight model into serious sales growth. On Friday, Bloomberg reported that the lab is targeting $2 billion in annualized revenue by the end of the year, double the company’s reported revenue run rate for August.
Why this matters
Moonshot's $2 billion target is a rounding error next to OpenAI and Anthropic's reported $40 billion runs, but the 300 billion daily tokens on OpenRouter tell us something more useful than any revenue projection: developers are actually routing production traffic through K3 at scale. That's a harder signal to fake than a fundraising target. For teams building on open-weight models, this is worth watching closely, because a Chinese lab hitting real usage numbers without US-style pricing or marketing muscle suggests the cost-performance calculus is shifting faster than the revenue gap implies.
The slight recent dip in usage is also worth noting, since it hints that Moonshot's growth curve may not be as linear as the Bloomberg-reported target assumes. We'd treat the $2 billion figure as aspirational until Moonshot posts actual quarterly numbers, not run-rate math. But the token volume is the part builders should track.
Revenue targets get revised. Token counts on a public router don't lie about who's actually shipping with a model.
Related Reading
Open Source D&B rebuilds 642 million‑business database after AI agents hit limits
Open Source OpenAI's GPT Transcribe Cuts Error Rate to 3.31%, Improving on GPT-4o
Open Source Study: 6.7% of Deepfake Requests on Hugging Face Targeted Children
From Archives Meta AI Suggested Invasive Question After User's Cross-Post
From Archives Cohere's 218B North Small Translate Model Scores 83.6 on WMT26
Common Questions Answered
How many tokens are K3 models processing daily on OpenRouter?
According to OpenRouter data, K3 models are currently processing as many as 300 billion tokens per day. This represents significant production traffic being routed through Moonshot AI's open-weight models at scale, even after usage cooled somewhat from its peak levels.
What is Moonshot AI's revenue target for the end of the year?
Moonshot AI is targeting $2 billion in annualized revenue by the end of the year, which represents a doubling of the company's reported revenue run rate from August. This aggressive growth target reflects the company's confidence in monetizing its popular Kimi chatbot and K3 models.
Why is the 300 billion daily token metric more significant than Moonshot's revenue projections?
The 300 billion daily tokens moving through K3 models on OpenRouter represents actual developer adoption and production traffic, which is a harder signal to fake than revenue projections or fundraising targets. This real usage data demonstrates that developers are genuinely routing their applications through K3 at scale, providing more concrete evidence of market traction than financial forecasts alone.
How does Moonshot AI's revenue target compare to other major AI companies?
While Moonshot AI's $2 billion annual revenue target is substantial, it is relatively small compared to OpenAI and Anthropic's reported $40 billion revenue run rates. However, the significance lies in Moonshot's achievement as a Chinese lab reaching real usage numbers at scale without relying on US-style funding models.
Further Reading
- China AI Star Moonshot Eyes $2 Billion Annualized Sales in 2026 - Bloomberg
- China's Moonshot AI Releases Kimi K3, Its Largest Open-Weight Model Yet - PureAI
- Kimi K3 Release: Signal vs Noise on Moonshot's 2.8T Model - Kie.ai
- Moonshot AI Releases Kimi K3, Its Largest Open-Weight Model Ever - Kimi.ai
- MoonshotAI/Kimi-K3: Open Frontier Intelligence - GitHub
