China's Kimi K3 AI model is now challenging premium frontier models with its performance. It joins Meta's Muse Spark and Google's Antigravity in offering startup founders and engineers sufficiently good performance for routine programming tasks at much lower cost than earlier favourites Fable and GPT.

Kimi is launching K3, a multimodal open-weight model with 2.8 trillion parameters and one million tokens of context. In the company's own benchmarks, it comes close to Claude…

Moonshot's Kimi K3 beat Claude Fable 5 and GPT 5.6 in some benchmarks... and it's an open model.

Moonshot AI's Kimi K3 is the largest open-weight model ever released - 2.8 trillion parameters, a million-token memory, benchmark scores within touching distance of GPT-5.6 Sol,…