Thinking Machines Lab unveils new Inkling version at 25pc size
Inkling-Small matches or exceeds Inkling on reasoning and agentic tasks, the company said. Read more: Thinking Machines Lab unveils new Inkling version at 25pc size
The Thinking Machines Lab has introduced a new open-weights model, Inkling-Small, which matches or exceeds the performance of its predecessor, Inkling, while being a quarter of its size. Developed in partnership with Nvidia, the model boasts 276 billion total parameters, with 12 billion being active. Inkling-Small outperforms Inkling on reasoning and agentic tasks, according to the company.
It has been rated 40% on the Intelligence Index, placing it "well above average" relative to other comparable models. The model also scored above 31% on Humanity's Last Exam, surpassing Inkling's 29.7%. Additionally, it achieved over 80% on SWEBench-Verified. Inkling-Small can process text, image, and audio inputs and produce text outputs, making it useful for real-world applications like cropping, zooming, and programmatic image inspection.
The lab aims to build AI systems that augment human will and judgment, allowing customers to develop highly customizable AI solutions that outperform closed models on various tasks faster and more affordably.
Written by urgent.news from Silicon Republic's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
