Inkling Small
Built for reasoning, agentic coding, and multimodal tasks where you want an open model that lets you control thinking effort.
Explore additional models
Frequently
asked questions
Inkling Small is the smaller, more efficient member of Thinking Machines Lab’s Inkling family of open-weight models. It is a mixture-of-experts transformer with 276B total parameters, 12B active, and it supports a 1M-token context on Crusoe. It accepts text, image, and audio input, ships under the Apache 2.0 license, and is built for reasoning, coding, and agentic workflows. It is available for high-performance inference via Self-Serve Deployments in Crusoe Intelligence Foundry.
Inkling Small FP8 Dynamic is an 8-bit version of Inkling Small, with the same architecture, parameter counts, context length, input types, and license. Its weights are stored in 8-bit precision, which roughly halves the GPU memory needed to serve it, with a small accuracy tradeoff. Choose Inkling Small for the most accurate outputs and Inkling Small FP8 Dynamic to serve on fewer GPUs.
Inkling Small is an open model built for reasoning and code. It suits scientific and technical reasoning, plus code generation tasks like writing code and solving programming problems.
Inkling Small lets you set thinking effort for each request, from minimal to extra-high. Use low effort for fast, high-volume tasks like extraction and classification, and higher effort for hard reasoning and agentic coding. Lower effort cuts latency and output tokens, so matching effort to the task keeps serving costs down.
Yes. You can deploy Inkling Small for high-performance inference via Self-Serve Deployments on any available GPU, with a profile tuned for throughput or responsiveness.


%204.png)
