Thinking Machines Lab releases open-weight Inkling 2 Mini
Original title: thinkingmachines/Inkling-2-Mini — new model trending #29 on Hugging Face
The Apache 2.0 checkpoint is about 20GB, with 26.4 billion parameters and 3.4 billion active per token.
Thinking Machines Lab released Inkling 2 Mini, an open-weight multimodal model that takes text and images and generates text. It is a 24-layer decoder-only transformer with looped middle layers and sparse mixture-of-experts routing, intended for agents, coding, chat and retrieval applications. Weights are on Hugging Face, with API access through the Tinker platform, and local inference is supported on listed NVIDIA, Apple and open-source runtimes.
Unconfirmed: The provided model card cuts off before comparative evaluation scores are shown.
- The release date is November 20, 2026, under Apache 2.0.
- Three loop passes yield about 5.4 billion effective active parameters per token.
- Each token routes to 8 of 128 experts; context is up to 1 million tokens.
- Recommended local hardware includes one NVIDIA DGX Spark or Apple silicon with 32GB or more.
- Vendors
- Thinking Machines Lab
- AI models
- Inkling 2 Mini
From the original article
5 min read at huggingface.coimage-text-to-text · 136 downloads · 70 likes · license apache-2. 0
By thinkingmachines · huggingface.co
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.