deberta-v3-small

deberta-v3-small is Microsoft’s lightweight text classification model optimized for fast, low-cost zero-shot classification. Provide text together with your own candidate labels, and the model returns confidence scores for each category without requiring a fixed taxonomy or task-specific training. It is well suited for intent detection, content tagging, routing, filtering, prioritization, and other high-volume classification workflows where speed and cost matter. Its compact architecture makes it a strong alternative to using a general-purpose LLM for straightforward classification tasks.