flywheel-ai/home-services

TEXT GENERATIONConcurrency Cost:3Model Size:35.1BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 20, 2026License:apache-2.0Architecture:Transformer Open Weights Cold

The flywheel-ai/home-services model is a 35.1 billion parameter vertical AI-employee model developed by Flywheel by OpSpot. Fine-tuned from Qwen/Qwen3.6-35B-A3B, it is specifically optimized for tasks within the home-services domain. This model leverages synthetic seed data from permissively-licensed local models, focusing on niche applications where its specialized knowledge provides an edge over general-purpose LLMs.

Loading preview...

Model Overview

flywheel-ai/home-services is a 35.1 billion parameter model developed by Flywheel by OpSpot, designed as a specialized "vertical AI-employee" for the home-services domain. It is fine-tuned (LoRA) from the Qwen/Qwen3.6-35B-A3B base model, which operates under an Apache-2.0 license.

Key Characteristics

  • Domain-Specific: Explicitly optimized for home-services related tasks and inquiries.
  • Provenance: Trained on synthetic seed data generated by permissively-licensed local models (Apache/MIT), ensuring no distillation from closed-source models.
  • Base Model: Built upon the robust Qwen3.6 architecture.
  • Accessibility: Available in multiple formats including safetensors for transformers and vLLM, and model-q4_k_m.gguf for llama.cpp and Ollama.

Deployment Options

This model supports flexible deployment across various platforms:

  • llama.cpp: Can be run using llama-server with the GGUF format.
  • Ollama: Directly pullable and runnable via ollama run hf.co/flywheel-ai/home-services.
  • vLLM: Serves the safetensors format for high-throughput inference.

Intended Use

This model is ideal for applications requiring deep, specialized knowledge within the home-services sector, where its fine-tuning provides a distinct advantage over more generalist large language models.

Finetunes
1 models