Small open typed-decision model with calibrated probabilities, first-class Nepali coverage.
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
choice, score, noul, classify, route
choice, classify, route
Architecture
mira
nativ
Fine-tuned from
jhu-clsp/mmbert-small
dicta-il/neodictabert
License
apache-2.0
cc-by-4.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
61.3%
96.2%
Calibration error
0.083
0.010
Valid action rate
100.0%
—
Median latency
30 ms
—
p95 latency
43 ms
—
Evaluation suite
s1-decision-bench
nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite)
Latest version
0.1.5
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 mira and nativ-he-decision?
mira is from SAGEA and nativ-he-decision from Yoav Pinto. Both have open weights you can download and run. Both answer choice, classify and route questions. Only mira answers score and noul. mira is licensed apache-2.0; nativ-he-decision, cc-by-4.0.
Which is more accurate, mira or nativ-he-decision?
They report on different suites — mira 61.3% on s1-decision-bench, nativ-he-decision 96.2% on nativ-bench user-intent task, 4 options (MASSIVE Hebrew, 2,973 items; the maker's own suite) — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, mira or nativ-he-decision?
mira: Free (open weights). nativ-he-decision: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run mira or nativ-he-decision locally?
Yes, both: systemone pull sagea/mira and systemone pull yoav-pinto/nativ-he-decision download the weights.
