level-ai/LevelField
LevelField is a 9-billion parameter multimodal model developed by Level AI, built upon Qwen3.5-9B. It specializes in interpreting 2D manufacturing drawings and generating optimized machining process plans. This model excels at extracting critical information from drawings and proposing operation sequences for non-standard machining, making it highly valuable for process engineers in manufacturing.
Loading preview...
LevelField: Multimodal AI for Manufacturing Process Planning
LevelField is a 9-billion parameter multimodal model developed by Level AI, designed to assist process engineers with non-standard machining process planning. Built on the Qwen3.5-9B base model, it interprets 2D manufacturing drawings (PDF and image documents) to draft optimized machining routes.
Key Capabilities
- Drawing Interpretation: Extracts essential details like part names, drawing numbers, materials, dimensions, tolerances, surface roughness, and technical requirements from engineering drawings.
- Process Planning: Proposes operation sequences, identifies unsuitable processes, and suggests improvements based on reviewed machining requirements.
- Extensive Process Coverage: Supports 18 distinct manufacturing process types, including turning, milling, drilling, casting, forging, heat treatment, and inspection.
- Multilingual Support: Operates in both Chinese and English.
Good For
- Automating Initial Process Route Drafting: Significantly reduces the time required to draft process routes, with reported reductions of approximately 60% in pilot customer usage.
- Assisting Process Engineers: Provides proposals that require engineering review, acting as a powerful assistant rather than a fully autonomous system.
- Interpreting Complex Drawings: Helps extract information from various drawing formats, though users should carefully check small annotations or ambiguous geometry.
LevelField was fine-tuned and reinforced using 100,000 manufacturing drawings and process sheets from 16 factories, with validation from 36 experienced process engineers. Internal evaluations report over 95% alignment between the model's process reasoning and senior engineers' decisions. It is released under the Apache 2.0 license, permitting commercial use.