Thinking Machines Lab launches Inkling 2 Small, a multimodal model on Tinker

Thinking Machines Lab released Inkling 2 Small, a mixture-of-experts model with 594 billion total parameters but only 17 billion active at a time, keeping it efficient despite its scale. It handles both images and text, supports a variable effort setting for trading off speed against quality, and can process up to 1 million tokens of context. On the Artificial Analysis Intelligence Index, it scores 31, more than double the 13 scored by the earlier Inkling Small. For now, the model is available only through the company's Tinker platform for inference and fine-tuning, with closed weights. Thinking Machines says it plans to release the weights under the permissive Apache 2.0 license next week, once external safety testing wraps up. The staged rollout, serving it first through a hosted platform and following with open weights, reflects a now-common pattern among labs balancing commercial access with eventual openness.

Why it matters

A meaningful jump in a mid-size open-weights model's intelligence score, with full open release expected within days.

Source: Thinking Machines Lab