Thinking Machines releases open Inkling 2 Mini looped model
Original title: New open model Inkling 2 Mini thinks in loops to solve problems with fewer words and still runs on a laptop
The Apache 2.0 model has 26.4 billion parameters, 5.4 billion active per token, and a 1-million-token window.
Thinking Machines released Inkling 2 Mini, an open mixture-of-experts looped transformer that reruns 12 of its 24 layers three times per token so it can reason longer without emitting more text. It handles text and images under Apache 2.0. The company says it leads open models under 100 billion parameters on agentic coding tests. Weights and a variable-loop research version are on Hugging Face, and it is available in the Tinker Playground.
Unconfirmed: Coding leadership and efficiency figures are company and Artificial Analysis claims, not independently verified here.
- It scores 48.2 percent on DeepSWE v1.1 and 46.5 percent on Vals Index v2.2.
- Artificial Analysis rates it 20, tying Qwen3.8 27B at lower cost and higher speed.
- It runs locally on Macs with at least 32 GB of memory and on Nvidia DGX Spark.
- Vendors
- Thinking Machines
- Organizations
- Artificial AnalysisHugging Face
From the original article
1 min read at the-decoder.com
Thinking Machines has released Inkling 2 Mini, an open looped transformer that runs 12 of its 24 layers three times per token. This lets the model reason internally for longer without generating extra text. On Artificial Analysis's Intelligence Index, it ties with Qwen3.
By Matthias Bastian · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.