FrontierThe story, in brief

Thinking Machines Lab Drops Its First Model

975B parameters. Thinking Machines just entered the arena with Inkling, an open-source multimodal model trained on video and audio.

Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

A new contender is establishing technical credibility in the crowded foundation model space with a multimodal approach, signaling intent to compete against well-funded incumbents.

The key facts

10 to know
  1. Inkling model: 975-billion parameters

  2. Open source release

  3. Multimodal training (video and audio)

  4. Thinking Machines Lab positioned against Anthropic and OpenAI

  5. Published July 15, 2026

  6. Thinking Machines Lab releases Inkling model

  7. 975 billion parameters

  8. Open source

  9. Multimodal: trained on video and audio

  10. Positioned as competitive alternative to Anthropic and OpenAI

Go to the source

Wired AIwired.com

Publisher excerpt: Inkling, a 975-billion-parameter open source model, was trained to understand video and audio. It could help Thinking Machines establish itself among competitors like Anthropic and OpenAI.
Read original report
Back to today's editionMore frontier news

Keep reading

Related stories

More from Frontier