Open-weights models now just 2 months behind closed frontier as Kimi K3.1 Pro lands
0xsachi · x · 2026-11-09
Sentient Foundation's Open-Source AI Field Notes (Nov 6, 2026) reports the open-source vs closed model gap has shrunk from years to just 2 months.
- Moonshot released Kimi K3.1 Pro, the best open-weights model for most coding, 3D modeling/CAD, and financial research tasks after Opus 5.5 and GPT-6 Astra, at lower cost. Benchmarks show it leading open models on SWE Marathon (51.7), Kimi Code Bench 3.0 Long-horizon (61.2), and Agents' Last Exam (43.2).
- Liquid AI shipped the LFM3 family of on-device specialized models (LFM3-1.7B/VL-2B/Audio-2B/Omni-2B, d1-Omni-2B), claiming parity with Gemma 4 12B, Gemini 3.5 Flash Lite, and GPT-5.6 Luna on less capable hardware.
- Artificial Analysis launched a new Progress Index v1; the strongest open-weights model, Kimi K3.1 Pro, now trails the closed frontier by only 2 months.
More from AGI Musings
- Andreessen: AI-native startups vs incumbents is 'widest gap' he's ever seen — ai · 2026-11-10
- OpenAI's Job Transition Framework adopted by Philadelphia Fed and World Bank to map AI job risk — soumitrashukla9 · 2026-11-10
- Vonnegut's 'buying one envelope' quote goes viral in AI circles: 'we're dancing animals' — msfeldstein · 2026-11-10
- Survey: a monitor-controller-mechanism framework for meta-reasoning in LRMs — kalyan_kpl · 2026-11-10
- Justine Bateman pushes back on tech bros' AI predictions: 'They can't predict our future' — GaryMarcus · 2026-11-10
- Environmental groups are using AI to fight wildfires and protect species — castrotech · 2026-11-09