The 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.
Native MLX FP16 conversion of Laya for Apple silicon. The same ModernBERT-large encoder, decision transformer and heads, running in MLX with no PyTorch; the checkpoint Laya Studio fine-tunes from.
Decides
choice, score, noul, classify, route
choice, score, noul
Architecture
decider
laya
Fine-tuned from
qwen/qwen3.5-2b-base
convai-innovations/laya
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
80.2%
—
Calibration error
0.038
—
Valid action rate
—
—
Median latency
3.2 ms
—
p95 latency
—
—
Evaluation suite
Decider 67-task regression set
Figures are from each model’s manifest; accuracy and latency are what the publishers report, on their own suites and hardware. Add a third model.
Questions
What is the difference between decider and laya-mlx?
decider is from Mapika and laya-mlx from aac6fef. Both have open weights you can download and run. Both answer choice, score and noul questions. Only decider answers classify and route. laya-mlx is the smaller model, at 421M parameters to 2.0B.
Which is more accurate, decider or laya-mlx?
Only decider publishes an accuracy figure (80.2% on Decider 67-task regression set); laya-mlx does not, so there is no comparison to make without your own test.
Which is cheaper, decider or laya-mlx?
decider: Free (open weights). laya-mlx: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run decider or laya-mlx locally?
Yes, both: systemone pull mapika/decider and systemone pull aac6fef/laya-mlx download the weights.
