About
OpenGCM is an independent, open-source AI research organization. Our main focus is Hydrion — a series of language models we're pretraining from scratch, aimed at understanding what small models can do when you build them yourself instead of just fine-tuning someone else's.
We also maintain GCM, a line of code-focused chat models built by fine-tuning existing open-weight models.
Everything we release — weights, training data, and methods — is public.
Hydrion
Our pretrained model series, trained from the ground up rather than fine-tuned.
| Model | Parameters | Type | Link |
|---|---|---|---|
| Hydrion v1 | 114M | Base | Hugging Face → |
| Hydrion v1-SFT | 114M | Instruction-tuned | Hugging Face → |
GCM
Code-focused chat models, fine-tuned on top of existing open-weight base models.
| Model | Parameters | Link |
|---|---|---|
| GCM Mark II | 9B | Hugging Face → |
GCM Mark II scores 74.4% pass@1 on HumanEval and 62.2% pass@1 on MBPP.
Vision
We believe AI should be:
- Open — transparent training data, weights, and methods
- Efficient — capable models that don't require massive infrastructure
- Customizable — built to be fine-tuned, extended, and adapted
- Accessible — usable by anyone, not gatekept behind closed systems
Research
Current focus areas:
- Small language model pretraining from scratch (Hydrion)
- Efficient fine-tuning of existing models (GCM)
- Model evaluation and benchmarking
- Local AI deployment
Follow along in our Research space.
Team
Built by MrGuineaBird, solhost, and contributors.