A 27B local model nearly beats DeepSeek's 1.6T flagship in Agent benchmark
A dark horse enters China's AI race: StartLux
StartLux, a new Chinese AI startup founded by internet veteran Chen Danyan, has surprised the industry with its first model, StartLux-V1.0-27B-Preview. In the CAICT MCP specialized test, this 27B-parameter local model scored 39.25%, ranking second only to DeepSeek-V4-Pro (1.6T parameters) by a mere 1.3 percentage points. The model runs on consumer PCs and excels in Agent tasks like financial analysis and browser automation, outperforming Claude Sonnet 4.6 in real-world scenarios. StartLux focuses on local model commercialization, challenging the cloud-based scaling paradigm.
Scaling Law is a 'passing fairy,' without it, AI cannot take off, but it has merely passed through the development path of AI and its future is not necessarily tied to it.
- RetroTechie
Anyone know how this company expects to make money from AI model(s) running locally on consumer-grade hardware?
Download a closed-source binary for <platform>, payment via appstore? Ads inserted in output? Freemium vs. pro tier offerings? Some other way?
This seems like boxed software of old, right? Cracked and distributed on warez sites in no-time.
- FailMore
I enjoyed this article... random cat pictures (esp. thumbs up cat picture) kept me going.
I do think on device AI will be a BIG thing... and basically change how we interact with our computers entirely - you will just ask to get things done. Low latency, free to use as much as you want. I think this future is hard to avoid/this will be a chunk of the future.
- ngl999
Still censored, no thanks.