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 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
choice, score, noul, classify, route
Architecture
cygnet
winnow
Fine-tuned from
google/gemma-4-12b-it
google/gemma-4-12b-it
License
mit
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
87.9%
85.7%
Calibration error
—
—
Valid action rate
—
—
Median latency
50 ms
143 ms
p95 latency
—
—
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 cygnet and winnow?
cygnet is from Blockbrain Labs and winnow from EldanRing. Both have open weights you can download and run. Both answer choice, score and noul questions. Only winnow answers classify and route. winnow reads up to 64K tokens of state, against 16K tokens for cygnet. cygnet is licensed mit; winnow, apache-2.0.
Which is more accurate, cygnet or winnow?
They report on different suites — cygnet 87.9% on JevBench public set (231 items), JevBench CLI, 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, cygnet or winnow?
cygnet: Free (open weights). winnow: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Which is faster, cygnet or winnow?
By their publishers’ figures, cygnet answers in about 50 ms at the median and winnow in about 143 ms — measured on different hardware, so treat it as a rough guide.
Can I run cygnet or winnow locally?
Yes, both: systemone pull blockbrain-labs/cygnet and systemone pull eldanring/winnow download the weights.
