Mira Murati’s Thinking Machines Launches Open-Weight AI Model ‘Inkling’

Thinking Machines Lab, the artificial intelligence startup founded by former OpenAI Chief Technology Officer Mira Murati, has unveiled its first in-house AI model, Inkling. The company describes Inkling as an open-weight large language model (LLM) capable of handling coding, reasoning and agentic tasks while allowing users to adjust its reasoning effort based on speed and performance preferences. Unlike most leading AI companies, Thinking Machines has released the model under an open-source licence, enabling developers and enterprises to download, modify and fine-tune it through platforms such as Hugging Face and the company's own Tinker customisation platform.

Inkling features a 975-billion-parameter Mixture-of-Experts (MoE) architecture, activating approximately 41 billion parameters for each task to improve efficiency. It supports a one-million-token context window and has been trained on 45 trillion tokens across text, images, audio and video, giving it native multimodal capabilities. The model was trained entirely on NVIDIA GB300 NVL72 systems. While Thinking Machines acknowledges that Inkling is not the strongest AI model currently available, it believes its openness, efficient reasoning and enterprise-focused customisation make it a strong foundation for future AI development. The company also introduced a lighter version, Inkling-Small, designed for lower cost and faster performance.