Inkling (batch)

Open-weight multimodal mixture-of-experts model with 41B active parameters from 975B total and a 524K token context window.

Strengths

Handles multimodal inputs, supports long-context reasoning with its 524K token window, and uses mixture-of-experts routing to activate only a portion of its parameters per request.

Best for

General-purpose reasoning tasks, coding, agentic workflows, and tool-use systems that benefit from larger capacity than Inkling Small.

Limitations

As an open-weight model, it requires self-hosting or managed deployment; the larger parameter count and context window increase computational requirements compared to Inkling Small.