OpenAI’s former CTO has launched a new company and open-sourced its first model, Inkling, with an architecture that draws inspiration from DeepSeek-V3

robot
Abstract generation in progress

According to Beating monitoring, Thinking Machines Lab, founded by OpenAI’s former CTO Mira Murati, has released its first general-purpose model, Inkling. The model has a total of 975 billion parameters, activates 41 billion per activation, supports text, images, and audio, and has a maximum context length of 1 million tokens.

Inkling is trained from scratch, but its MoE architecture is mainly based on DeepSeek-V3. In the early stages of post-training, it also used data generated by open models such as Kimi K2.5. The complete weights have been uploaded to Hugging Face, using the Apache 2.0 license.

Thinking Machines has clearly acknowledged that Inkling is not currently the strongest model. It focuses on customization and efficiency. When achieving the same results on Terminal Bench 2.1, the number of tokens used is about one-third of NVIDIA Nemotron 3 Ultra.

NVDA0.33%
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned