Thinking Machines Lab released Inkling, a 975-billion-parameter open-weights multimodal model under the Apache 2.0 license, on July 15, 2026. The model is a natively multimodal Mixture-of-Experts system with 975 billion total parameters and 41 billion active parameters, built to reason across text, images, and audio. It can be downloaded, modified, and run on-premises or in a VPC, positioned as a Western alternative with low-cost, customizable, open-weight AI. VentureBeat emphasized censorship resistance, while TechCrunch stressed customization via Tinker. Thinking Machines says Inkling is "not the strongest model available today, closed or open," reporting 77.6% on SWE-bench Verified and 91.4% on VoiceBench. VentureBeat reported Inkling scores 77.6% on SWE-bench Verified, beating Nvidia Nemotron 3's 71.9%, and posts 97.1% on AIME 2026 while Nemotron 3 is at 94.2%. The company also previewed Inkling-Small, a 276-billion-parameter alternative optimized for low latency and cost, with weights on Hugging Face and through its Tinker API.