Inkling 2 Mini: 26.4B-parameter open model runs offline on a 32GB MacBook

Thinking Machines Lab released Inkling 2 Mini on November 20, 2026, under an Apache 2.0 license. The 26.4-billion-parameter model fits in about 20GB of memory and can run offline on a 32GB MacBook. It uses a looped transformer design that reprocesses its middle layers, giving each token more reasoning depth without storing extra weights. The company reports benchmark results competitive with larger open-weight models, including 79.2% on SWE-bench Verified. The model quickly became the top trending model on Hugging Face.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)
Log in to join the discussion and vote.
Log in