An open Jev-style decision model by Andy Zhang: a LoRA adapter on Gemma 4 26B-A4B (also on Gemma 4 12B) that reads a state and returns a temperature-calibrated probability for every option of a yes/no, choice or score question in one forward pass.
Small open typed-decision model with calibrated probabilities, first-class Nepali coverage.
Decides
choice, score, noul
choice, score, noul, classify, route
Architecture
ajev
mira
Fine-tuned from
google/gemma-4-26b-a4b-it
jhu-clsp/mmbert-small
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
—
61.3%
Calibration error
—
0.083
Valid action rate
—
100.0%
Median latency
49 ms
30 ms
p95 latency
—
43 ms
Evaluation suite
Sequential requests, the maker's own measurement
s1-decision-bench
Latest version
2026.09
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 ajev and mira?
ajev is from Andy Zhang and mira from SAGEA. Both have open weights you can download and run. Both answer choice, score and noul questions. Only mira answers classify and route.
Which is more accurate, ajev or mira?
Only mira publishes an accuracy figure (61.3% on s1-decision-bench); ajev does not, so there is no comparison to make without your own test.
Which is cheaper, ajev or mira?
ajev: Free (open weights). mira: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Which is faster, ajev or mira?
By their publishers’ figures, mira answers in about 30 ms at the median and ajev in about 49 ms — measured on different hardware, so treat it as a rough guide.
Can I run ajev or mira locally?
Yes, both: systemone pull andy-zhang/ajev and systemone pull sagea/mira download the weights.