Moonshot AI’s Kimi K3 tops coding benchmark

Beijing-based Moonshot AI has released Kimi K3, a 2.8 trillion parameter model it describes as the world’s first open 3T-class system and the largest open-weight AI model to date. Moonshot said the model still trails Anthropic’s Claude Fable 5 and OpenAI’s GPT 5.6 Sol overall, but beat other models in its evaluation suite across coding and agentic benchmarks.

Kimi K3 has a 1 million token context window, native vision, and activates just 16 of its 896 experts per token, or about 1.8% of the pool. Frontend Code Arena ranked it first with 1,679 points in blind developer testing, ahead of Claude Fable 5. API pricing is listed at $0.30 per million cache-hit input tokens, $3 per million on cache misses, and $15 per million output tokens.

Moonshot attributed performance gains to Kimi Delta Attention, Attention Residuals, and quantization-aware training using MXFP4 weights and MXFP8 activations. Its blog references Nvidia’s H200, Nvidia L20, and an unnamed alternative GPU vendor, underscoring how Chinese labs are navigating export constraints. Published results remain unverified until the weights are made public on July 27.