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.
The most-downloaded open System One reproduction. Merged Qwen3.5 fine-tunes, trained on about 95 public decision sets, then calibration-aware RL and a rank-64 LoRA, that softmax letter logits at an answer slot.
Decides
choice, score, noul
choice, score, noul, classify, route
Architecture
cygnet
decider
Fine-tuned from
google/gemma-4-12b-it
qwen/qwen3.5-2b-base
License
mit
apache-2.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
87.9%
80.2%
Calibration error
—
0.038
Valid action rate
—
—
Median latency
50 ms
3.2 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 decider?
cygnet is from Blockbrain Labs and decider from Mapika. Both have open weights you can download and run. Both answer choice, score and noul questions. Only decider answers classify and route. decider is the smaller model, at 2.0B parameters to 12B. cygnet is licensed mit; decider, apache-2.0.
Which is more accurate, cygnet or decider?
They report on different suites — cygnet 87.9% on JevBench public set (231 items), JevBench CLI, decider 80.2% on Decider 67-task regression set — so the numbers do not rank them. Test both on your own labelled examples.
Which is cheaper, cygnet or decider?
cygnet: Free (open weights). decider: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Which is faster, cygnet or decider?
By their publishers’ figures, decider answers in about 3.2 ms at the median and cygnet in about 50 ms — measured on different hardware, so treat it as a rough guide.
Can I run cygnet or decider locally?
Yes, both: systemone pull blockbrain-labs/cygnet and systemone pull mapika/decider download the weights.
