Inkling-Small
Thinking Machines Lab · 2026-07-30 · 266B parameters
Inkling-Small is Thinking Machines Lab's open-weight multimodal Mixture-of-Experts model. Its model card reports 276 billion architectural parameters and 12 billion active parameters, while the released safetensor artifacts contain 265,956,439,090 tensor parameters. The 42-layer checkpoint accepts text, image, and audio, supports controllable reasoning effort and a 1,048,576-token context window, and supports BF16, MXFP8, and NVFP4 inference numerics.