ReOpenly

[ReOpenly @reopenly_ai]

A small language model is the same transformer architecture, sized to run where your data already lives: 1B to 15B parameters, not hundreds of billions.

No official cutoff exists. The working one: it fits on a single GPU.

[ReOpenly @reopenly_ai]

NVIDIA is acquiring Hugging Face. Open weights just got the deepest compute in the industry.

Our job does not change: portability. Every model you fine-tune on ReOpenly exports as open weights, with its adapter and config.

Good tools let you leave!

[ReOpenly @reopenly_ai]

That's impressive!

Finetuned Qwen3.5-0.8B beats GPT 5.6 Sol xhigh in specialized tasks!

Small models is definitely the future.
This is not a random person attesting this, this is the CEO of Shopify.

[tobi lutke @tobi]

Training tiny models for special purpose use cases works so incredibly well if you have a great self improving recursive flywheel. Shopify ML team is on fire.

finetuned 0.8b model beats GPT 5.6-sol xhigh in this very specialized task.

[ReOpenly @reopenly_ai]

Introducing https://t.co/6YSrpHiyeI 🚀

The end-to-end platform for small language models. Tuned to your task, a small open model matches frontier accuracy at a fraction of the cost.

Own your intelligence: private, compatible, no lock-in.
Ever.

Small models, frontier results.