flywheel-ai/legal-intake

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 24, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

flywheel-ai/legal-intake is a 35.1 billion parameter vertical AI model fine-tuned from Qwen/Qwen3.6-35B-A3B by Flywheel by OpSpot. This specialized model functions as a professional law-firm intake coordinator, excelling at tasks like new-client intake, matter categorization, conflict-check fact capture, and scheduling. It is specifically designed for the legal-intake domain, providing focused capabilities beyond general-purpose LLMs.

Loading preview...

Overview

flywheel-ai/legal-intake is a specialized 35.1 billion parameter AI model developed by Flywheel by OpSpot. It is fine-tuned using LoRA from the Qwen/Qwen3.6-35B-A3B base model, licensed under Apache-2.0. This model is engineered to act as a dedicated vertical AI-employee within the legal sector, specifically for intake processes.

Key Capabilities

This model is designed to perform the functions of a professional law-firm intake coordinator, including:

  • New-client intake: Efficiently gathering initial client information.
  • Matter categorization: Classifying legal matters appropriately.
  • Conflict-check fact capture: Documenting necessary details for conflict resolution.
  • Document-collection checklists: Generating lists of required documents.
  • Consult scheduling: Assisting with the arrangement of client consultations.
  • Fee and retainer logistics: Handling initial financial discussions.
  • Status updates: Providing information on case progression.

Provenance and Guardrails

The model's v1.0 was trained on synthetic seed data generated by permissively-licensed local models (Apache/MIT). It is explicitly not a lawyer and is programmed to never give legal advice; instead, it captures legal questions to route them to licensed attorneys. The model is available in safetensors format for transformers/vLLM and model-q4_k_m.gguf for llama.cpp/Ollama.