NewsBite analysis · AI, Startups
TechCrunch
Thinking Machines amps up its bet against one-size-fits-all AI with its first open model, Inkling
Thinking Machines Lab has unveiled its first proprietary AI model, Inkling, which features 975 billion parameters and enables organizations to customize the model according to their needs. This launch challenges traditional AI model approaches from major companies by promoting adaptability over cookie-cutter solutions. Inkling, trained on diverse media inputs, aims for efficiency and relevance in enterprise applications.
Act nowConfidence: HighImmediate horizon
Published 15 Jul 2026, 18:40 UTC6 min read
Visit original article ↗Lead image — TechCrunch
Summary
TL;DR
- Inkling has 975 billion total parameters but uses only 41 billion for specific tasks.
- The model is trained on 45 trillion tokens across text, audio, and visual data.
- Thinking Machines positions Inkling as a starting point for enterprise AI customization.
- The model emphasizes calibrated responses and efficient resource usage.
- Inkling's performance is designed to be adjustable based on organizational needs.
Interesting facts
- Inkling is marketed as an enterprise product, diverging from conventional one-size-fits-all models.
- Its development was completed in about nine months, significantly faster than OpenAI’s timeline.