decision-models/baseten-groundhog-latest
A long-context, multimodal mixture-of-experts decision model from Base Labs
groundhog-1 is a decision model post-trained from Gemma 4 26B A4B for classification, routing, and scoring. Give it text or images and a set of questions; it returns structured decisions and option probabilities without generating free-form text. Its 131,072-token context window supports decisions over long inputs.
| Attribute | Value |
|---|---|
| Parameter Size | 26B-A4B |
| Context Length | 128k |
| Input Modalities | Text, Image |
| Output Modalities | Structured Decisions |
Use it for tasks such as routing requests, classifying cases, answering yes/no checks, or scoring against a rubric. It supports choice questions, yes/no questions, and ordered scoring, returning probabilities your application can use to select an answer or set confidence thresholds.
curl --fail-with-body \
-X POST "https://inference.baseten.co/v1/systemone" \
-H "Authorization: Api-Key $BASETEN_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "decision-models/baseten-groundhog-latest",
"state": {
"message": "I was charged twice. Please refund the duplicate charge."
},
"questions": {
"route": {
"type": "choice",
"instructions": "Which team should handle this request?",
"criteria": {
"billing": "Payments, charges, and refunds",
"support": "Technical troubleshooting"
}
}
}
}'