Moonshot AI has launched its Kimi K3 model, a 2.8-trillion-parameter Mixture-of-Experts architecture with a 1 million token context window. The model is positioned as a cost-effective alternative to competitors like GPT-5.6 and Claude Fable 5, offering significant savings on API usage, particularly for high-volume applications. Kimi K3 also features a novel prompt caching mechanism that further reduces costs for repeated contexts, and access is available through the Yingsuan AI unified gateway, which offers compatibility with the OpenAI SDK. AI
IMPACT Sets a new price-performance benchmark for large context window models, potentially driving down costs for AI applications.
RANK_REASON Frontier-lab model release with system card and benchmark data. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →