Laya fine-tune that scans untrusted text (extracted files, knowledge-base documents, agent skills, tool descriptions) for prompt injection and data exfiltration in English and German, as a noul question answered in one forward pass. 322M parameters, PyTorch and ONNX.
A 364M Hebrew decision encoder on NeoDictaBERT: give it a state, a question and 2 to 7 free-text options and it returns a probability for each option in one forward pass, about 70 ms on a laptop CPU. Hebrew only.
Decides
noul, classify
choice, classify, route
Architecture
laya
nativ
Fine-tuned from
convaiinnovations/laya-multilingual
dicta-il/neodictabert
License
Unspecified; training data includes CC BY-NC-SA 4.0 material (non-commercial), not relicensed
cc-by-4.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
—
96.2%
Calibration error
—
0.010
Valid action rate
—
—
Median latency
95.4 ms
—
p95 latency
149 ms
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 laya-cybersec and nativ-he-decision?
laya-cybersec is from TextCortex and nativ-he-decision from Yoav Pinto. Both have open weights you can download and run. Both answer classify questions. Only laya-cybersec answers noul. Only nativ-he-decision answers choice and route. laya-cybersec is the smaller model, at 322M parameters to 364M. laya-cybersec is licensed other; nativ-he-decision, cc-by-4.0.
Which is more accurate, laya-cybersec or nativ-he-decision?
Only nativ-he-decision publishes an accuracy figure (96.2% on nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite)); laya-cybersec does not, so there is no comparison to make without your own test.
Which is cheaper, laya-cybersec or nativ-he-decision?
laya-cybersec: Free (open weights). nativ-he-decision: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run laya-cybersec or nativ-he-decision locally?
Yes, both: systemone pull textcortex/laya-cybersec and systemone pull yoav-pinto/nativ-he-decision download the weights.
—
Evaluation suite
TextCortex saved run: 279 single-window inputs, PyTorch batch one
nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite)
