Moonshot Kimi K3 pushes Chinese AI into Fable level territory, undercutting Anthropic and OpenAI on price
Category: Foundation Models • Competition
What it is
Moonshot AI released Kimi K3, a 2.7-trillion-parameter open-weight model it calls
its most powerful coding model yet, now the largest open-weight LLM available,
ahead of DeepSeek V4. Moonshot claims K3 performs competitively with
Anthropic’s Fable 5 and substantially outperforms Opus 4.8, GPT-5.6 Sol, and
GPT-5.5 on its own published benchmarks. K3 costs $15 per million output tokens- pricier than China’s GLM-5.2 and DeepSeek V4, but far below Fable 5’s $50.
Analysts hadn’t expected a Fable-level Chinese model before next year.
Why it Matters for Enterprises
The gap between US and Chinese frontier models is closing faster than expected.
Enterprises benchmarking AI vendors should evaluate open-weight Chinese
alternatives on capability and total cost, not just brand familiarity.
