flywheel-ai/construction
The flywheel-ai/construction model is a 35.1 billion parameter language model developed by Flywheel by OpSpot, fine-tuned from Qwen/Qwen3.6-35B-A3B. This model is specifically optimized as a vertical AI-employee for the construction domain, excelling as a practical assistant for tasks like estimating, materials, code awareness, and scheduling. It supports a 32768 token context length and is available in safetensors and GGUF formats.
Loading preview...
Overview
Flywheel-ai/construction is a 35.1 billion parameter language model developed by Flywheel by OpSpot. It is fine-tuned (LoRA) from the Qwen/Qwen3.6-35B-A3B base model, specifically designed as a vertical AI-employee for the construction industry. This model aims to serve as a practical assistant for various construction and trades-related tasks.
Key Capabilities
- Estimating and Takeoffs: Assists with calculating project costs and material quantities.
- Materials and Methods: Provides information on construction materials and techniques.
- Code and Permit Awareness: Offers guidance on building codes and permit requirements.
- Scheduling and Sequencing: Helps with planning project timelines and work order.
- Jobsite Safety: Provides insights into safety protocols and best practices.
Provenance and Usage
This model (v1.0) was trained using synthetic seed data generated by permissively-licensed local models (Apache/MIT teachers only), ensuring no distillation from closed-source models. While its general performance is comparable to its base model, its specialized capabilities are enhanced for niche construction-related queries. It is available in safetensors for transformers/vLLM and model-q4_k_m.gguf for llama.cpp/Ollama.