Contrastive Language Models score a state against a set of candidate actions with a contrastive objective. Two projection heads on frozen Qwen3-8B embeddings, trained with InfoNCE; clm-serve maps Choice, Score and Noul onto candidate ranking.
StartLux's decision family in five sizes, 0.8B to 27B. All questions in a request are answered in one forward pass, with a probability for every option, through a TypeSafe /v1/systemone-compatible server that records CUDA graphs for short requests.
Decides
choice, score, noul, rank, classify, route
choice, score, noul
Architecture
clm
startlux-decision
Fine-tuned from
qwen/qwen3-8b
—
License
apache-2.0
cc-by-nc-4.0
Availability
Open weights
Open weights
Hosted by
—
—
Input price
—
—
Decision accuracy
—
88.3%
Calibration error
—
—
Valid action rate
—
—
Median latency
—
26 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 clm and startlux-decision?
clm is from Contrastive-LM and startlux-decision from StartLux. Both have open weights you can download and run. Both answer choice, score and noul questions. Only clm answers rank, classify and route. startlux-decision is the smaller model, at 4.7B parameters to 8.0B. clm is licensed apache-2.0; startlux-decision, cc-by-nc-4.0.
Which is more accurate, clm or startlux-decision?
Only startlux-decision publishes an accuracy figure (88.3% on JevBench public set (231 items; 204 correct), maker's run); clm does not, so there is no comparison to make without your own test.
Which is cheaper, clm or startlux-decision?
clm: Free (open weights). startlux-decision: Free (open weights). Open weights cost nothing per call beyond your own hardware.
Can I run clm or startlux-decision locally?
Yes, both: systemone pull contrastive-lm/clm and systemone pull startlux/startlux-decision download the weights.
—
Evaluation suite
—
JevBench public set (231 items; 204 correct), maker's run
