flywheel-ai/home-services

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jun 20, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

Flywheel by OpSpot developed flywheel-ai/home-services, a 35 billion parameter model fine-tuned from Qwen/Qwen3.6-35B-A3B. This specialized model functions as a vertical AI-employee, specifically optimized for the home-services domain. It excels at tasks like quoting, scheduling, maintenance planning, and basic troubleshooting for HVAC, plumbing, and electrical contractors. The model is trained on synthetic seed data from permissively-licensed local models.

Loading preview...

Overview

Flywheel-ai/home-services is a specialized 35 billion parameter language model developed by Flywheel by OpSpot. It is fine-tuned (LoRA) from the Qwen/Qwen3.6-35B-A3B base model, licensed under Apache-2.0, and is designed as a "vertical AI-employee" for the home-services industry.

Key Capabilities

  • Home-Services Specialization: Optimized for tasks relevant to HVAC, plumbing, and electrical contractors.
  • Practical Assistance: Provides support for quoting, estimates, scheduling, dispatch, maintenance plans, parts and equipment inquiries, warranty questions, and service-call triage.
  • Basic Troubleshooting: Offers homeowner-level troubleshooting guidance.
  • Model Formats: Available in safetensors (for transformers/vLLM, ~65GB) and model-q4_k_m.gguf (for llama.cpp/Ollama, ~20GB).

Provenance & Guardrails

The model's v1.0 is trained on synthetic seed data derived from permissively-licensed local models, ensuring no distillation from closed-source models. It includes a critical guardrail: it is not a substitute for licensed professionals and defers to them or emergency services for gas leaks, electrical hazards, water damage, or structural/permit-regulated work.

Finetunes
1 models