China’s Moonshot AI and U.S. startup Thinking Machines each released major open-weight models this week, pushing the open-source AI field into direct competition with frontier proprietary systems. The releases arrive as open-weight models claim 41% of Hugging Face downloads and dominate the top six slots on OpenRouter — all from Chinese firms — signaling a structural shift in how production AI gets deployed.
Moonshot AI Releases Kimi K3, a 2.8-Trillion-Parameter Open Model
Moonshot AI’s Kimi K3 is now the largest open-source AI model ever released, with 2.8 trillion total parameters — roughly 75% larger than DeepSeek’s V4 Pro at approximately 1.6 trillion parameters, according to Moonshot’s technical documentation. The Beijing-based startup, backed by Alibaba, timed the release to precede the 2026 World Artificial Intelligence Conference in Shanghai. Benchmark results place Kimi K3 neck-and-neck with Anthropic’s and OpenAI’s most capable proprietary systems.
The model features a 1-million-token context window and native visual understanding. Full model weights are scheduled for release on July 27, per details reviewed by researchers who examined the company’s technical documentation. The model is already accessible via kimi.com with a Google account or phone number — no credit card required. The release marks a notable recovery for Moonshot AI, whose market position had eroded significantly over the prior 18 months following DeepSeek’s rise.
Thinking Machines Launches Inkling Under Apache 2.0
Thinking Machines — the American AI startup founded by former OpenAI CTO Mira Murati — released Inkling, its first open-weights model, under an Apache 2.0 license on Thursday. The model is a 975-billion-parameter Mixture-of-Experts system capable of reasoning across text, images, and audio.
Inkling scores 77.6% on SWE-bench Verified, outperforming NVIDIA’s Nemotron 3 at 71.9%, and 91.4% on VoiceBench, compared to 94.4% for Gemini 3.1 Pro on high reasoning effort, according to Thinking Machines’ release announcement. The company also flagged that Inkling was designed “to answer directly on topics that may be subject to censorship,
Related news
- you show me kimi k3 is not benchmaxxed, i cancel my claude subscription right now and i go work with open-weights – Reddit Singularity
- Dario addresses the Kimi K3 situation – Reddit Singularity
Sources
- Thinking Machines open sources first multimodal language model, Inkling, focused on low cost and ‘resistance to censorship’ – VentureBeat
- China’s Moonshot AI releases Kimi K3, the largest open-source model ever, rivaling top U.S. systems – VentureBeat
- Fine-Tuning Your First Large Language Model (LLM) with PyTorch and Hugging Face – HuggingFace Blog
- RAG vs Fine-Tuning Explained: What They Actually Do and When to Use Each – Towards Data Science
- The real AI race may no longer be at the frontier – TechCrunch
