featherless-ai/Qwen3.8-27B-classifier

Model Overview

The featherless-ai/Qwen3.8-27B-classifier is a 27 billion parameter model built on the Qwen architecture, specialized for classification tasks via the Simple Jev classifier API. Unlike general-purpose LLMs, this model is engineered to convert diverse inputs—including text, chat history, and images—into structured, actionable JSON outputs. It is particularly useful for applications requiring precise decision-making based on contextual information.

Key Capabilities

• Structured Decision Output: Provides JSON responses for direct application integration, including chosen candidates, confidence scores, and probabilities.
• Versatile Question Types: Supports three primary classification types:
  • choice: Selects from named candidates (e.g., routing support tickets).
  • score: Rates items on an ordered rubric (e.g., urgency or relevance).
  • noul: Evaluates yes/no propositions with a confidence score.
• Multi-modal Context: Processes text, structured data, chat histories, and supports image input, making it suitable for visual classification tasks.
• Optimized for Classification: Focuses on scoring model logits for defined answer labels, providing a robust framework for various classification needs.

Good For

• Intent Detection: Identifying user intent from queries or conversations.
• Content Moderation: Classifying content based on predefined criteria.
• Support Routing: Directing customer inquiries to appropriate departments.
• Relevance Scoring: Evaluating the relevance of information or documents.
• Automated Decision-Making: Generating structured decisions for agent actions or workflow automation.