ComfyUI Extension: LLM Text Processor

ComfyUI-LLM-text-processor Introduction

The ComfyUI-LLM-text-processor is an extension designed to enhance the capabilities of ComfyUI by integrating local Large Language Models (LLMs) using the llama.cpp framework. This extension allows you to process both text and images with various GGUF LLMs, including popular models like Qwen3-VL, Qwen3.5, Qwen3.6, Gemma 4, and gpt-oss. It provides a local LLM node within ComfyUI for tasks such as prompt writing, rewriting, translation, captioning, and text extraction. By leveraging local models, this extension offers a powerful tool for AI artists to generate and transform text, as well as to work with multimodal models that combine text and image inputs.

How ComfyUI-LLM-text-processor Works

At its core, the ComfyUI-LLM-text-processor works by integrating local LLMs into the ComfyUI environment. It discovers models stored in the ComfyUI/models/LLM directory and uses them to perform various text processing tasks. The extension can handle both text and image inputs, making it versatile for different creative workflows. For multimodal models, it uses an external mmproj file to process images alongside text. This setup allows AI artists to seamlessly incorporate advanced language models into their creative processes without needing extensive technical knowledge.

ComfyUI-LLM-text-processor Features

• Text Generation and Transformation: Utilize local GGUF models to generate and transform text, enhancing your creative projects with AI-generated content.
• Multimodal Support: Process images alongside text using llama.cpp models that support multimodal inputs, including handling batches of images.
• Customizable Outputs: Separate outputs for RESPONSE and REASONING, allowing you to tailor the output to your specific needs.
• System Prompt Presets: Use text files to create system prompt presets, streamlining the setup for different tasks.
• Automatic Model Discovery: The extension automatically discovers models in the specified directory, simplifying the setup process.
• Advanced Options: For users who need more control, advanced llama.cpp options are available, including memory management and processing toggles.

ComfyUI-LLM-text-processor Models

The extension supports a variety of model families, each suited for different tasks:

• Qwen Models: Ideal for text and vision tasks, with variants like Qwen3-VL, Qwen3.5, and Qwen3.6.
• Gemma 4 Models: Known for strong text generation capabilities, with support for multimodal tasks depending on the specific GGUF release.
• gpt-oss Models: Offers options for both direct answers and reasoning-heavy tasks, with models like gpt-oss-20b and gpt-oss-120b. Each model family has specific strengths, and choosing the right one depends on your project's requirements, such as the need for reasoning or multimodal capabilities.

Troubleshooting ComfyUI-LLM-text-processor

Here are some common issues and solutions:

• No Models Appear: Ensure that at least one .gguf model is placed in the ComfyUI/models/LLM directory and restart ComfyUI.
• Image Input Fails: Verify that the mmproj file is correctly set and matches the model family of the selected GGUF model.
• llama.cpp Setup Fails: Check your internet connection and GitHub access, then try running the node again.
• Out of Memory: Reduce the ctx_size or use a smaller model to manage memory usage effectively.
• Slow Generation: Lower the max_tokens or ctx_size, or consider using a smaller model to speed up processing.

Learn More about ComfyUI-LLM-text-processor

To further explore the capabilities of the ComfyUI-LLM-text-processor, consider visiting the following resources:

• ComfyUI GitHub Repository
• llama.cpp GitHub Repository
• Qwen Documentation
• Gemma Model Overview
• gpt-oss Documentation These resources provide additional insights and support for using the extension effectively in your creative projects.