All models

COMPARE

2 models, side by side.

What each one decides, how well calibrated it is, how fast it answers and what it costs to pull. Up to 4 at a time; the better value in each row is marked.

Propertymapika/deciderwfzyx/von
SummaryThe most-downloaded open System One reproduction. Merged Qwen3.5 fine-tunes, trained on about 95 public decision sets, then calibration-aware RL and a rank-64 LoRA, that softmax letter logits at an answer slot.A ModernBERT-large encoder with an option-marker head. Premise and options are packed into one sequence and each option's marker is scored in a single bidirectional pass; version 1.2 makes the scoring order-invariant.
Decideschoice, score, noul, classify, routechoice, score, noul, classify
Architecturedecidervon
Fine-tuned fromqwen/qwen3.5-2b-baseanswerdotai/modernbert-large
Licenseapache-2.0apache-2.0
AvailabilityOpen weightsOpen weights
Hosted by——
Input price——
Decision accuracy80.2%63.9%
Calibration error0.0380.045
Valid action rate——
Median latency3.2 ms18 ms
p95 latency——
Evaluation suiteDecider 67-task regression setJevBench public standard tier
Latest version2026.091.2.0
Variants——
Size of latest version3.5 GB2.9 GB
Files42
Downloads00
Stars00
Tagssystem-one, qwen, rl, calibrated, 2bsystem-one, encoder, modernbert, 395m
UpdatedSep 25, 2026Sep 25, 2026