Julia 1
A 144M-parameter System One model on the multilingual mmBERT-small encoder. Takes a state, a question and 2 to 20 options and returns one decision with probabilities for choice, score and yes/no questions, on a CPU.
Supersonic Labs reports 73.15% on the typed-decisions test set (400 cases, 2,000 questions), against 72.70% for Jev on the same set: choice 71.3%, yes/no 80.7%, score 68.9%. On MASSIVE scenario classification it reaches 71.5% across 52 locales. The runtime accepts 8,192 combined tokens (the benchmarks used 1,024) and 2 to 20 options of up to 48 tokens each. It loads through the repository's own Python package rather than Transformers; an ONNX and WebGPU export is published as SupersonicLabs/Julia-1-ONNX.
What it decides
- choice — picks one option from a set
- score — places the input on an ordered scale
- noul — answers a yes/no question with one calibrated probability
- classify — assigns a category from a fixed taxonomy
- route — sends the input to one of several destinations
At a glance
| Parameters | 144M |
| Base model | jhu-clsp/mmbert-small |
| Maker | Supersonic Labs |
| Released | 2026-09-23 |
| License | apache-2.0 |
| Reported accuracy | 73.2% |
Get the weights
pip install systemonemodels
systemone pull supersonic-labs/julia-1
The files are served from the maker's Hugging Face repository, SupersonicLabs/Julia-1, and verified against the checksums recorded here.
Read more
This page was opened by System One for Supersonic Labs, who can claim the organisation and take it over at any time.