Thinking Machines 发布 Inkling 2 Mini:26B 参数可本地运行
Thinking Machines 把 594B 的 Inkling 2 Small 压缩成 26B 的 Mini,笔记本就能跑,AA Index 还能拿 20.1 分。想在本地跑智能体但没大卡的朋友可以直接试试,比同级别的 Llama 系小模型在 agentic 任务上更有看头。
Thinking Machines 发布 Inkling 2 Mini,一个仅 26B 参数的 looped transformer 模型,可在笔记本上本地运行。此前不少用户反映 Inkling 2 Small 的 594B 参数规模过大,Mini 是对这一反馈的回应。模型在 AA Index 基准上拿到 20.1 分,官方称其足以胜任 agentic 工作负载。相比 594B 的 Inkling 2 Small,Mini 把参数量压缩到不足其二十分之一,同时保持 looped transformer 架构带来的推理能力,定位是本地部署的小型前沿模型。对于需要在本地跑智能体的开发者,这是一个无需大规模算力的选择。
值得注意的是,官方将其描述为"西方少有的可本地运行的前沿模型",开源社区对本地部署小模型的需求正在推动厂商推出更小的衍生版本。从 594B 到 26B 的产品线布局,说明 Thinking Machines 在用同一架构覆盖不同算力场景。
A state of the art model from the West that you can run locally. Very timely Let’s go 🔥 Thinking Machines @thinkymachines Some of you thought that Inkling 2 Small was too big at 594B parameters. We heard you: Inkling 2 Mini is a looped transformer with just 26B parameters, small enough to run on your laptop, powerful enough for agentic work with a score of 20.1 on AA Index. thinkingmachines.ai/news/inkling-2… 🔗 View Quoted Tweet📊 7 ⚡
This report is categorized under AITOP's "Models" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Thinking Machines, Inkling 2 Mini, open-source models, agents.
Source attribution: curated from Julien Chaumond (@julien_c).