Former OpenAI CTO Murati's Thinking Machines Releases Open-Weights 975-Billion-Parameter Model
Sourced from 4 publications
- •Thinking Machines, founded by former OpenAI CTO Mira Murati, released Inkling as an open-weights model with 975 billion parameters.
- •Inkling is multimodal, trained natively on video and audio rather than being limited to text processing.
- •The Register positions the release as a deliberate contrast to both OpenAI's closed models and the growing field of Chinese open-weights LLMs.
- •TechCrunch describes the launch as Thinking Machines' first public proof after eighteen months of behind-the-scenes infrastructure development.
- •Open weights differs from open source: model parameters are public, but training data and underlying code are not.
What Happens Next
- →The 975B parameter count limits practical deployment to well-resourced enterprises and cloud providers with high-end GPU clusters, concentrating Inkling's near-term impact among hyperscalers and large research labs rather than the broader open-source community.
- →Inkling's native video and audio training creates a new open-weights benchmark for multimodal AI, accelerating development of video analytics, media production, and audio-processing applications by removing the need for proprietary multimodal APIs from OpenAI or Google.
- →OpenAI faces a targeted reputational challenge from its former CTO publicly championing open weights, intensifying internal and external pressure on OpenAI to release more model artifacts or risk losing developer mindshare to credible alternatives led by ex-insiders.
Near-term: Cloud providers (AWS, Azure, GCP) race to offer hosted Inkling inference endpoints within weeks, using Murati's model as a differentiation lever against competitors, while distillation efforts begin to produce smaller derivatives accessible on consumer hardware. Long-term: The precedent of a credible ex-OpenAI executive succeeding with open weights normalizes the open-weights business model for frontier-scale AI, fragmenting the market away from the closed-API oligopoly and pushing regulatory frameworks to distinguish between open-weights and full open-source licensing.
Sources
Curated from 4 sources. Every summary is reviewed for accuracy, but may still contain errors. We always link to original sources for verification.
Related Stories
About Meridian
Meridian is a free daily newsletter delivering signal-scored news stories with forward-looking analysis every morning. Stories are scored across six criteria (global leverage, capital impact, temporal durability, career relevance, decision utility, and narrative clarity) then assigned to Big Signal, Core, or Quick tiers.
Get Meridian in your inbox
The stories that matter, every morning at 06:00.