Thinking Machines Lab drops its first model, Inkling

Wired reports Thinking Machines Lab — the closely watched startup founded by former OpenAI CTO Mira Murati — released its first model, Inkling, entering the frontier landscape at a moment of intense lab competition. The debut is significant less for immediate benchmark supremacy than for what it signals: Murati's lab, backed by one of the largest seed rounds in history, is now shipping product rather than operating in stealth.

The release process itself drew commentary from within the company; Murati amplified a note that 'it truly takes a village to release a model, perhaps especially an open weights model,' suggesting Inkling follows the open-weights path that Kimi K3, Thinking Machines, and others are championing this week. That aligns Thinking Machines with the open-source moment that Yann LeCun and Clement Delangue argued this week will define the frontier's next phase.

Strategically, a first model from a lab of this pedigree is a credibility milestone — it converts hype and funding into a testable artifact. But the bar is brutal: Inkling debuts into a field where Claude Fable 5, GPT-5.6 Sol, and now open 3T-class Kimi K3 already set high marks, and a first model rarely leads on raw capability.

The open question is differentiation. Thinking Machines has emphasized research on model behavior and reliability; whether Inkling embodies a distinctive approach — or is a competent-but-conventional first release — will determine how seriously the market takes it. Watch for independent benchmarks, the licensing terms of the weights, and whether the lab articulates a clear technical thesis that separates it from the crowded frontier.