According to @godofprompt, Chinese AI startup Moonshot has released the Kimi K2 model, a 1 trillion-parameter AI trained for $4.6 million, significantly less than the billions spent by US labs on models like GPT-5. Kimi K2 outperformed OpenAI’s flagship on critical benchmarks, achieving 44.9% on 'humanity’s last exam' compared to proprietary models, and leading in agentic browsing tasks with 60.2% versus GPT-5’s 54.9%. The model executes 200-300 tool calls autonomously, highlighting advancements in reasoning and automation. Unlike many closed US models, Kimi K2 is open-source under a modified MIT license, with 32B active parameters per token, native int4 quantization for double speed, and a 256k context window, making it accessible for commercial AI applications on affordable hardware. This launch demonstrates a shift in the AI race, showing that rapid deployment and open access can rival, or even surpass, high-budget proprietary efforts, creating new business opportunities for AI-driven products and services (source: @godofprompt, Nov 10, 2025).

According to DeepLearning.AI, Beijing-based Moonshot AI has launched the Kimi K2 LLM family, providing open-weights access to a groundbreaking one trillion-parameter language model under a modified MIT license. The fine-tuned Kimi-K2-Instruct model achieved 53 percent on LiveCodeBench and 76.5 percent on AceBench, outperforming existing large language models in code generation and reasoning benchmarks. This development enables broader adoption and innovation in generative AI, offering significant business opportunities for enterprises seeking advanced AI-powered solutions and fostering increased collaboration within the global AI ecosystem (source: DeepLearning.AI, July 29, 2025).