Project Gaia

An open research project: a small AI that knows UK pet-care rules properly — and knows when to say “ask your vet”.

Project Gaia is a research project by MyPetSitter Ltd, Harrogate, UK. It asks whether a small, specialised language model, trained only on material it is allowed to use, can be more reliable on this one subject than a large general one — answering from the records it is given, citing them, carrying their cautions and vet referrals, and saying so when the records do not answer. It also asks whether a vocabulary and data mix trained jointly for English and Romanian gives Romanian a fair share of a model's capacity. Round 1 covers English and Romanian.

The models were trained from scratch on Isambard-AI, under UKRI's AI Research Resource (AIRR) award.

Website: https://projectgaia.co.uk

Round 1 — released 12 November 2026

releasewhat it is
modelgaia-3b-chatThe Round 1 chat model, version 1.0: 3 billion parameters, English and Romanian, 131,072-token context, fine-tuned to answer pet-care questions from numbered records in its prompt, citing them, and to say when they do not answer. A research model: read its “Known limits and risks” first.
modelgaia-3b-base-128kThe 3B base model it is made from: pre-trained from scratch on 150 billion tokens of openly licensed English, Romanian, pet-care, code and maths text, then extended to a 131,072-token context. It continues text; it is not chat-tuned.
datasetgaia-records-benchThe Gaia Records Benchmark v0.1: held-out pet-care and animal-law questions in English and Romanian, answered from numbered official records; it measures citing the right record, carrying its cautions and vet referrals, and saying when the records do not answer.

Each model card says how every figure was measured and who read it, and ships with the data card that names every source of training data and its licence.

Licence

How to cite

The benchmark (from its CITATION.cff):

@misc{petrisor2026gaiarecordsbench,
  author = {Petrisor, Marian V.},
  title  = {Gaia Records Benchmark: pet care and animal law, English--Romanian},
  year   = {2026},
  note   = {Version 0.1.0, released 2026-11-12},
  url    = {https://huggingface.co/datasets/ProjectGaiaUK/gaia-records-bench}
}

The models:

@misc{petrisor2026gaia3b,
  author = {Petrisor, Marian V.},
  title  = {Project Gaia Round 1: gaia-3b-chat and gaia-3b-base-128k},
  year   = {2026},
  note   = {Round 1, version 1.0, released 2026-11-12. Apache-2.0},
  url    = {https://huggingface.co/ProjectGaiaUK}
}

And Isambard-AI, as its acknowledgement asks: McIntosh-Smith, S., Alam S. R. and Woods, C. (2024). “Isambard-AI: a leadership class supercomputer optimised specifically for Artificial Intelligence”. https://doi.org/10.48550/arXiv.2410.11199

Acknowledgement

The authors acknowledge the use of resources provided by the Isambard-AI National AI Research Resource (AIRR). Isambard-AI is operated by the University of Bristol and is funded by the UK Government’s Department for Science, Innovation and Technology (DSIT) via UK Research and Innovation; and the Science and Technology Facilities Council [ST/AIRR/I-A-I/1023].

UKRI AI Research Resource (AIRR) Gateway · project 0261-1548-0489-1 · 4,000 GPU hours of computing time.

The award supports the research. It is not an endorsement of any product.

Cite: McIntosh-Smith, S., Alam S. R. and Woods, C. (2024). “Isambard-AI: a leadership class supercomputer optimised specifically for Artificial Intelligence”. https://doi.org/10.48550/arXiv.2410.11199


Project Gaia · MyPetSitter Ltd, UK · https://projectgaia.co.uk