@indigox: Finally, the US also has a fully open-weight large model, after the failure of Llama... Thinking Machines' first OpenWeight Model - Inkling, trained entirely from scratch on GB300s, with 975 billion parameters, 41 billion active...

X AI KOLs Timeline Models
open-weight multimodal large-language-model fine-tuning reasoning

Summary

Thinking Machines has released Inkling, a fully open-weight multimodal large model with 975 billion parameters, supporting text, images, and audio. It has 41 billion active parameters and a 1 million token context window, and can be fine-tuned on Tinker.

Finally, the US also has a fully open-weight large model, after the failure of Llama... Thinking Machines' first OpenWeight Model - Inkling, trained entirely from scratch on GB300s, with 975 billion parameters, 41 billion active, 1 million context, fully multimodal (text, image, audio), controllable reasoning depth. There's something new to play with! The more diverse the models, the better for users 😘
Original Article
View Cached Full Text

Cached at: 07/16/26, 08:14 AM

Finally, the US also has a fully open-weight model, after Llama fell short… Thinking Machines’ first OpenWeight Model - Inkling, trained from scratch entirely on GB300s, with 975 billion parameters, 41 billion active, 1 million context window, fully multimodal (text, image, audio), controllable reasoning depth — something new for everyone to play with! The more diverse the models, the better for users.

Only Gemini supports video.

Similar Articles

Inkling: Our Open-Weights Model

Hacker News Top

Thinking Machines AI releases Inkling, a new open-weights mixture-of-experts multimodal foundation model with 975B total parameters and 41B active, supporting text, images, audio, and video, along with a preview of Inkling-Small.

@Phoenixyin13: This is the brand new best open-source model from the US. Thinking Machines Lab directly released a 975B parameter MoE multimodal giant, fully open-source under Apache-2.0! On standard benchmarks, it even outperforms NVIDIA's Nemotron strong model. Its core...

X AI KOLs Timeline

Thinking Machines Lab released a new 975B parameter MoE multimodal open-source model under Apache-2.0 license, surpassing NVIDIA's Nemotron on standard benchmarks, supporting text, image, and audio modalities with only 41B activated parameters, efficient and deployment-friendly.