Skip to content

AI.Model

Name Mandatory Description Default Type
⬅️ Input The input of the shard, if any String
Output ➡️ The resulting output of the shard LLMModel
ISQ No In-situ quantization bit width. Quantizes at load time. ISQBits::None ISQBits
Files No GGUF filename(s) within the repo. When set, uses GGUF loader. none StringNone
UQFF No UQFF filename (e.g. 'q4k-0.uqff'). When set, loads pre-quantized UQFF model — no ISQ needed. none StringNone
Embedding No When true, loads as an embedding model for use with AI.Embed. false Bool

Load a model via mistral.rs. Accepts a HuggingFace model ID or local path. Auto-detects architecture. For GGUF models, set the Files parameter. For embedding models, set Embedding: true.