Thinking Machines Launches Inkling 2 Small, a Sparse Architecture Aimed at Enterprise Agent Tasks

Original sourceVentureBeat

Summary

Thinking Machines released Inkling 2 Small, the second generation of its Inkling model, positioned near frontier performance on enterprise benchmarks. The model is a 594B-parameter sparse architecture activating only about 17B parameters per token, available for inference and customization via the…

Key points

For enterprises that want to build their own agents while retaining a self-hosting option, this is a new choice in sparse large models and licensing terms.
A sparse architecture plus an open-weights commitment could lower inference costs and vendor lock-in risk for enterprise agent deployments.
When evaluating agent platforms, enterprises can factor in per-token active parameters, self-hosting, and licensing terms in cost and compliance comparisons.

AI 模型

Editorial note

This page is Code & Chain's editorial summary of public sources. It may be prepared with AI assistance and published through an automated workflow. Refer to the original sources; this content is not investment, legal, or tax advice.