this-that-model-1.2
FLock.io's typed-decision model, fine-tuned from Mapika's decider-2b. Reads one hidden state per question and scores the declared options, so an answer outside the list cannot occur; several questions about one state share a forward pass.
Version 1.2 (23 September) targets policy rules however they are phrased, including rules written in the negative, and rules that compose. On FLock's public 1,710-question complex-decision benchmark it scores 0.878 (1.1: 0.775, 1.0: 0.406) and 0.909 on a held-out composed-rule ladder. Up to 255 options. Released under MIT, adapted from the Apache-2.0 decider-2b, which this registry already lists; the pip package is thisthat. The third-party Decision Index 0.2.1 scores 1.2 at 28.14.
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
At a glance
| Parameters | 1.88B |
| Base model | mapika/decider-2b |
| Maker | FLock.io |
| Released | 2026-09-20 |
| License | mit |
| Reported accuracy | 87.8% |
| Reported latency | 31.4 ms per question on an RTX 5080 Laptop GPU |
Get the weights
pip install systemonemodels
systemone pull flock-io/this-that-model
The files are served from the maker's Hugging Face repository, flock-io/this-that-model-1.2, and verified against the checksums recorded here.