
Thinking Machines releases Inkling 2 Mini for local use on laptops
The AMW Read
The open-weight release adds a laptop-scale model to a known lab's lineup and advances competition over model size and local deployment.
Thinking Machines releases Inkling 2 Mini for local use on laptops
Thinking Machines Lab released Inkling 2 Mini on November 20, 2026. The model has 26.4 billion total parameters, with 3.4 billion active per token, and accepts text and images while producing text. Its weights are available under an Apache 2.0 license, and the model requires about 20 GB of memory. Thinking Machines also offers it for inference and customization through its Tinker platform. The company says it leads open models under 100 billion parameters on agentic coding tests; the reported results are company claims.
The release adds a laptop-scale open-weight option to the foundation-model market. It follows Inkling 2 Small, a 594-billion-parameter model that was available at launch only through Tinker. The Mini uses a looped transformer that repeats two middle blocks of layers three times per token, reusing weights while increasing the effective activation count. This design makes the tradeoff between model size and computation relevant alongside the choice between local deployment and a hosted service.
Builders evaluating the Mini can download its weights or use Tinker with an account. Local use requires about 20 GB of free memory, so laptop compatibility depends on the available memory. The source also describes an oversight question: repeated internal computation can lengthen work performed without a written chain of thought. Teams should evaluate performance and monitoring needs for their own tasks rather than rely on the company's coding benchmark claim alone.



