Typed decisions from Google's frozen Gemma-4-12B-it, with no fine-tuning. A small shim over unmodified vLLM letters the options, reads the model's own probability for each letter at one answer position and applies one calibration temperature; one output token per decision.
A rank-16 LoRA on Qwen3.5-4B built for the JevBench setting. A document, a policy and a question go in; a calibrated distribution over the option letters comes out of one forward pass. Research and demo use only.
Decides
choice, score, noul
choice, score, noul, classify
Architecture
cygnet
hopper
Fine-tuned from
google/gemma-4-12b-it
qwen/qwen3.5-4b
License
mit
Research and demo use only (training data includes RACE, non-commercial); serving code Apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
87.9%
68.5%
Calibration error
—
0.102
Valid action rate
—
—
Median latency
50 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 cygnet and hopper?
cygnet is from Blockbrain Labs and hopper from HopitAI. Both have open weights you can download and run. Both answer choice, score and noul questions. Only hopper answers classify. hopper is the smaller model, at 4.0B parameters to 12B. cygnet is licensed mit; hopper, other.
Which is more accurate, cygnet or hopper?
They report on different suites — cygnet 87.9% on JevBench public set (231 items), JevBench CLI, hopper 68.5% on JevBench public hard tier (111 items, measured by the authors) — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, cygnet or hopper?
cygnet: Free (open weights). hopper: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run cygnet or hopper locally?
Yes, both: systemone pull blockbrain-labs/cygnet and systemone pull hopit-ai/hopper download the weights.
Evaluation suite
JevBench public set (231 items), JevBench CLI
JevBench public hard tier (111 items, measured by the authors)
