webAI 开源 TwiL-LM3，一款 1.7B 参数的正式逻辑推理模型，在 5 项形式推理基准中 4 项超越 OpenAI 的 GPT-OSS-120B，参数量少 40 倍，推理速度快 2.6 倍，可在手机和树莓派等消费级硬件上运行。该模型通过 289MB LoRA 适配器微调基础模型而来，基于专有数据引擎训练，专注于将自然语言转化为形式逻辑并验证结论。
webAI built a 1-gigabyte logic expert open-source model that runs on a phone and out-reasons models more than twice its size
TwiL-LM3, a 1.7B formal-logic model that beats OpenAI's gpt-oss-120b on 4 of 5 formal reasoning benchmarks while running efficiently on consumer hardware. Available on Huggingface.
That's 40× fewer parameters, 2.6× faster inference
The model does one specialized task, turning plain English into formal logic and checking whether conclusions follow from premises.
i.e. taking statements written in ordinary language and working out, step by step, what does and does not follow from them.
Its trained on a purpose-engineered, proprietary formal-logic data engine built from open sources.
Instead of training from scratch, webAI fine-tuned base models with a 289MB LoRA adapter of roughly 72M parameters.
Today, we’re excited to open-source TwiL-LM3, the first formal reasoning model from the webAI Intelligence Lab. At just 3 billion parameters, TwiL-LM3 outperforms OpenAI’s GPT-OSS-120B on 4 of 5 formal reasoning benchmarks while running efficiently on consumer hardware. That’s 40× fewer parameters, 2.6× faster inference, and state-of-the-art performance in the reasoning tasks that power reliable tool calling, code generation, structured outputs, and AI agents. TwiL-LM3 was trained using webAI’s proprietary reasoning pipeline on webAI-owned, verified datasets—not scraped internet data. We believe better reasoning comes from better training pipelines and higher-quality data, not simply larger models. Our approach demonstrates that efficient models can rival—and in many cases surpass—models dozens of times their size. Designed for the edge, TwiL-LM3 runs on hardware people already own—from a Raspberry Pi to an iPhone—bringing advanced reasoning to millions of devices without relying on the cloud. This is our first open-source release from the webAI Intelligence Lab, and it’s only the beginning. Proudly built in Austin, Texas. Article: https://www.webai.com/blog/webai-releases-twil-lm-a-family-of-formal-logic-models-that-outreason-a-120b-model-and-run-on-an-iphone在 X 查看被引用的帖子
来源：Rohan Paul · x.com