An open System One model from Interfaze AI. A rank-32 LoRA on Qwen3.5-4B reads yes/no, choice and score answers from the logits of one forward pass, returns calibrated probabilities and serves the /v1/systemone API.
A Gemma 4 12B fine-tune for typed decisions that also chats and reads images. Ships as GGUF for llama.cpp, holds a 64K context on a 16 GB GPU, and serves /v1/systemone next to /v1/chat/completions.
Decides
choice, score, noul, classify, route
choice, score, noul, classify, route
Architecture
lev
winnow
Fine-tuned from
qwen/qwen3.5-4b
google/gemma-4-12b-it
License
apache-2.0
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
68.9%
85.7%
Calibration error
0.115
—
Valid action rate
—
—
Median latency
69 ms
143 ms
p95 latency
—
—
Evaluation suite
S1Bench (13 public subsets, macro)
JevBench public subset (231 items), Q8_0
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 lev and winnow?
lev is from Interfaze AI and winnow from EldanRing. Both have open weights you can download and run. Both answer choice, score, noul, classify and route questions. lev is the smaller model, at 4.0B parameters to 12B.
Which is more accurate, lev or winnow?
They report on different suites — lev 68.9% on S1Bench (13 public subsets, macro), winnow 85.7% on JevBench public subset (231 items), Q8_0 — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, lev or winnow?
lev: Free (open weights). winnow: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Which is faster, lev or winnow?
By their publishers’ figures, lev answers in about 69 ms at the median and winnow in about 143 ms — measured on different hardware, so treat it as a rough guide.
Can I run lev or winnow locally?
Yes, both: systemone pull interfaze-ai/lev and systemone pull eldanring/winnow download the weights.