Moonshot's Kimi K2 Thinking Emerges as Leading Open Source AI, Outperforming GPT-5, Claude Sonnet 4.5 on Key Benchmarks

Why It Matters

The breakthrough shows that open‑source AI can match or exceed the capabilities of costly proprietary models, giving enterprises a high‑performance, low‑cost alternative and intensifying competitive pressure on major AI vendors.

Summary

Moonshot AI unveiled Kimi K2 Thinking, a trillion‑parameter Mixture‑of‑Experts model that activates 32 billion parameters per inference and is released under a modified MIT license on Hugging Face. In third‑party benchmarks it posted 44.9% on Humanity’s Last Exam, 60.2% on BrowseComp, 71.3% on SWE‑Bench Verified and 83.1% on LiveCodeBench v6, outpacing OpenAI’s GPT‑5, Anthropic’s Claude Sonnet 4.5 and xAI’s Grok‑4. The model is freely accessible via Moonshot’s platform and APIs, with usage priced at $0.15‑$2.50 per million tokens, far below proprietary alternatives. Its open‑weight release marks the first time an open‑source system has overtaken leading closed models on high‑end reasoning and coding tasks.