AI Models on 梦兽编程https://rexai.top/en/categories/ai-models/Recent content in AI Models on 梦兽编程Hugo -- 0.163.3en梦兽编程Sat, 01 Aug 2026 09:30:00 +08002.78 Trillion Parameters, 29GB RAM, Half a Token Per Second: Kimi K3 Actually Runs on a Laptophttps://rexai.top/en/posts/2026-08-01-waste-kimi-k3/Sat, 01 Aug 2026 09:30:00 +0800https://rexai.top/en/posts/2026-08-01-waste-kimi-k3/WASTE, a pure-C inference engine, squeezes the 2.78-trillion-parameter Kimi K3 into a 64GB laptop: a 982GB container streams expert weights from disk on demand, measured at 0.5 tok/s. Half a token per second is absurdly slow — but it proves running trillion-scale models locally is an engineering problem, not a money problem.Claude Opus 5 Underrated, Kimi K3 1.56TB Open-Weight, GLM 5.2 Becomes Security Default: July 2026 Model Landscape Shifthttps://rexai.top/en/posts/2026-07-29-model-landscape-2026/Tue, 28 Jul 2026 23:01:00 +0800https://rexai.top/en/posts/2026-07-29-model-landscape-2026/July's triple hit: Anthropic's Opus 5 tops coding but benchmarks underrate it, Moonshot's 2.8T-param Kimi K3 goes open-weight with a stricter license, and Zhipu's GLM 5.2 chosen by HuggingFace as their forensic tool — three forces reshaping the open/closed model landscape.