Thinking Machines, the AI startup founded by former OpenAI CTO Mira Murati, released Inkling, an open-weight model that anyone can download, run locally, and fine-tune, as the company's first major public release since its high-profile, heavily funded launch. The choice of format is the notable part: an executive who spent years building some of the most closely guarded closed models in the industry chose to debut her independent company with weights anyone can inspect and modify, rather than an API-only product behind a paywall. Inkling joins a fast-growing list of serious open-weight options that already includes DeepSeek, Qwen, GLM, and Kimi, and its arrival the same week Kimi K3 promised frontier-scale open weights reinforces a trend that is no longer confined to Chinese labs: giving away the model and building a business around services, hosting, fine-tuning, and enterprise support on top of it, the same playbook Mistral has run in Europe. For developers evaluating which models to self-host or fine-tune for a specific domain, Inkling is a new entrant worth testing directly against the existing open-weight field rather than assuming it inherits either the strengths or the limitations of Murati's prior work, since its actual architecture and training approach are Thinking Machines' own. The deeper business question the release raises, and does not answer, is what Thinking Machines charges for if the model itself is free: the likely answer, based on how other labs have handled the same trade-off, is enterprise deployment support, custom fine-tuning services, and hosted infrastructure rather than per-token API fees on the base model.