iCreatAI News
Skip to content
The Decoder· Matthias Bastian·· 13 hr agoSame storyAI score78

New open model Inkling 2 Mini thinks in loops to solve problems with fewer words and still runs on a laptop

New open model Inkling 2 Mini thinks in loops to solve problems with fewer words and still runs on a laptop

AI summary

Thinking Machines released Inkling 2 Mini, an open Apache 2.0 looped transformer mixture-of-experts model with 26.4 billion parameters (5.4 billion active per token) that runs 12 of its 24 layers three times per token so it can reason internally longer without generating more text. It handles text and images with a 1-million-token context window, leads open models under 100 billion parameters on agentic coding tests (48.2 percent on DeepSWE v1.1, 46.5 percent on Vals Index v2.2), and scores 20 on Artificial Analysis's Intelligence Index, tying Qwen3.8 27B at roughly six times lower cost and four times the speed. It runs locally on Macs with at least 32 GB of memory and on Nvidia's DGX Spark, with weights and a variable-loop research version on Hugging Face and a demo in the Tinker Playground.

Featured coverage of this story: 《Thinking Machines has released Inkling 2 Mini, an open weights looped transformer》

Source: The Decoder · the-decoder.com