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.
Berget AI's System One model for gating agent commands: a LoRA adapter and a fine-tuned joint schema head on Cloudflare's Clef-Flash that answer noul, choice and score questions over a state in one forward pass. Trained on Swedish and English operations decisions.
Decides
choice, score, noul
noul, choice, score
Architecture
ajev
clef
Fine-tuned from
google/gemma-4-26b-a4b-it
cloudflare/clef-flash
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
—
97.0%
Calibration error
—
—
Valid action rate
—
—
Median latency
49 ms
—
p95 latency
—
—
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 bev?
ajev is from Andy Zhang and bev from Berget AI. Both have open weights you can download and run. Both answer choice, score and noul questions. bev is the smaller model, at 9.0B parameters to 26B.
Which is more accurate, ajev or bev?
Only bev publishes an accuracy figure (97.0% on Berget held-out risk split (16,902 questions; same operations-traffic corpora as training)); ajev does not, so there is no comparison to make without your own test.
Which is cheaper, ajev or bev?
ajev: Free (open weights). bev: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run ajev or bev locally?
Yes, both: systemone pull andy-zhang/ajev and systemone pull berget-ai/bev download the weights.
Evaluation suite
Sequential requests, the maker's own measurement
Berget held-out risk split (16,902 questions; same operations-traffic corpora as training)
