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.
OpenAI's decision endpoint, announced at DevDay 2026. It focuses GPT-6 Luna on developer-defined questions that have a finite set of predefined answers, with text or image context, for classification, routing and an agent's next action.
Decides
choice, score, noul, rank, classify, route
choice, classify, route
Architecture
clm
gpt-6-luna
Fine-tuned from
qwen/qwen3-8b
—
License
apache-2.0
proprietary
Availability
Open weights
Hosted API
Hosted by
—
OpenAI
Input price
—
—
Decision accuracy
—
—
Calibration error
—
—
Valid action rate
—
—
Median latency
—
—
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 decisions-api?
clm is from Contrastive-LM and decisions-api from OpenAI. clm has open weights you can download and run; decisions-api is only available as a hosted API. Both answer choice, classify and route questions. Only clm answers score, noul and rank. clm is licensed apache-2.0; decisions-api, proprietary.
Which is more accurate, clm or decisions-api?
Neither publishes an accuracy figure. Test both on your own labelled examples.
Which is cheaper, clm or decisions-api?
clm: Free (open weights). decisions-api: Hosted, price not published. Open weights cost nothing per call beyond your own hardware.
Can I run clm or decisions-api locally?
clm yes — systemone pull contrastive-lm/clm downloads its weights. The other is only served as a hosted API.
