An independent Jev-shaped reproduction on DeBERTa-v3-large. One state and any number of choice, score and yes/no questions go in, and a calibrated distribution per question comes out of one forward pass. Public gold labels only.
A 4B decision model fine-tuned from Qwen3.5-4B-Base. A state and typed questions (choice, yes/no, score) go in; a temperature-calibrated probability for every option comes out of one pass, through a self-hosted /v1/systemone-compatible server. Apache-2.0 weights and code.
Decides
choice, score, noul, classify
choice, score, noul, classify, route
Architecture
open-jev
wald
Fine-tuned from
microsoft/deberta-v3-large
qwen/qwen3.5-4b-base
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
85.4%
88.3%
Calibration error
0.022
0.045
Valid action rate
—
—
Median latency
28 ms
33 ms
p95 latency
—
168 ms
Evaluation suite
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 open-jev-deberta-v3-large and wald?
open-jev-deberta-v3-large is from Kotoba Labs and wald from ORG2 AI. Both have open weights you can download and run. Both answer choice, score, noul and classify questions. Only wald answers route. wald reads up to 128K tokens of state, against 512 tokens for open-jev-deberta-v3-large. open-jev-deberta-v3-large is the smaller model, at 434M parameters to 4.2B.
Which is more accurate, open-jev-deberta-v3-large or wald?
They report on different suites — open-jev-deberta-v3-large 85.4% on Kotoba held-out test, seen question types (banking77, SST-5, BoolQ), wald 88.3% on JevBench public set (231 items; 204 correct), maker's run with the JevBench CLI, v1.2, effort none; items used as a development scoreboard — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, open-jev-deberta-v3-large or wald?
open-jev-deberta-v3-large: Free (open weights). wald: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Which is faster, open-jev-deberta-v3-large or wald?
By their publishers’ figures, open-jev-deberta-v3-large answers in about 28 ms at the median and wald in about 33 ms — measured on different hardware, so treat it as a rough guide.
Can I run open-jev-deberta-v3-large or wald locally?
Yes, both: systemone pull kotoba-labs/open-jev-deberta-v3-large and systemone pull org2ai/wald download the weights.
Kotoba held-out test, seen question types (banking77, SST-5, BoolQ)
JevBench public set (231 items; 204 correct), maker's run with the JevBench CLI, v1.2, effort none; items used as a development scoreboard
Latest version
2026.09
1.2.0
Variants
—
Dockerfile, LICENSE, NOTICE, model-files, reference, server
