North ML

Small AI. Serious research.

North ML is an independent AI research lab focused on advancing the frontier of language models through efficient, high-impact research.

We build research models that punch above their weight — exploring how far careful architecture, training, data, and fine-tuning can push a model beyond what its size or compute budget would suggest.

Our work spans small and efficient models, experimental architectures, fine-tuning, distillation, training methods, and evaluation.

Our goals

Advance the frontier.
Explore new ideas and build models that challenge what is expected from their scale.

Punch above the weight.
Maximize capability per parameter, per token, and per unit of compute.

Make research accessible.
Investigate approaches that can be developed and run without hyperscale infrastructure.

Publish openly.
Share models, datasets, experiments, and failures whenever possible so others can learn from and build upon our work.

We believe frontier AI isn't only about making models bigger.

It's about making them better..)

Model Series

• Starlight Mini (one model, discontinued series)
• Wind series (performed well, but it went.)
• Forge series (Possible return near the end of 2026, if I feel like it.)
• Aurora series (done.)
• Ember series (done.)
• North series (🤫, Air, Star and Arc. The long-awaited sequel may return.)

Compute Info

• M4 Mac Mini (16 gb variant, Jul 2025)
• M4 MacBook Air (16 gb variant, Dec 2025)
• v6e-16 tpu (Google Tpu research cloud, Feb 27, 2026 - April 31, 2026)
• Rtx Pro 6000 (Google Cloud, Janurary/Feburary - May 2026, then Modal)
• 8x L4 (Trained Aurora One, Modal)
• L40S (Modal)
• Other Modal GPUs.
• 5060 Ti (Jul 2026) 96

Announcement (May 7, 2026 AD, 3:52 PM)

A new release is expected on May 8, possibly Wind Arc 1.7, Wind Edge 1.7, or something new altogether. Thank you for using our models. I hope you enjoy what’s next.

Announcement (May 10, 2026 AD, 9:59 PM)

Thanks to our friends at Modal for giving us ~$180 in free credit to use B200, L4, and other GPUs.