Load Model Name in config.ini

If you're using more than one LLM provider in this pack and you're tired of retyping base_url and api_key into every single LLM node on the graph, this is the node that fixes that. It's a dropdown, not a text box - and the difference matters.

comfyui_LLM_party supports a genuinely long list of providers through a single OpenAI-compatible interface: OpenAI itself, Ollama, Azure OpenAI, llama.cpp, Grok, Qwen, GLM, DeepSeek, Kimi/Moonshot, Doubao, Gemini, and more - the README lists them under "Model support." Rather than pasting the same base_url/api_key combination into every LLM node that needs a given provider, the pack lets you configure each model once in the project's config.ini file, under its own name. This node is how you pull one of those names back out onto the graph: pick it from a dropdown, and it outputs the model name as a string you feed into your LLM node's model_name input, while the actual connection details stay centralized in one config file instead of scattered across every node that uses that model.

The one input. model_name - an enum, so it's a dropdown populated from what's actually configured in your config.ini, not free text. The default sample values you'll see out of the box include placeholders and provider examples like deepseek-chat, glm-4, qwen-max, moonshot-v1-8k, doubao_YOUR_ENDPOINT_ID, your_ollama_model_name, your_azure_openai_model_name, grok-beta, and gemini-1.5-flash - these map directly to the providers the README documents support. The output is model_name, a STRING, which plugs into any LLM node in this pack expecting a model name.

Setting up config.ini. This node is only as useful as what's actually in that file. Open config.ini in the comfyui_LLM_party project folder, and for each model you use regularly, add an entry with its base_url and api_key (or, for Ollama, base_url set to http://127.0.0.1:11434/v1/ with api_key set to ollama) under that model's name - config.ini.example in the same folder shows the format. Once that's done, this node's dropdown reflects it, and any LLM node downstream just needs the name, not the credentials.

Installing it. Load Model Name comes with the full comfyui_LLM_party pack - no separate install. Get it via ComfyUI Manager (search "comfyui_LLM_party") or cd ComfyUI/custom_nodes && git clone https://github.com/heshengtao/comfyui_LLM_party, then pip install -r requirements.txt from inside the project folder using your ComfyUI Python environment, and restart.

Where this actually trips people up. If the dropdown shows nothing but the placeholder, that's config.ini not being configured yet - this node reads from that file, so an empty or default config gives you an empty or default dropdown. Second, this node only outputs the name - it doesn't validate that the model behind it is reachable, so a wrong base_url or an expired api_key in config.ini won't surface here; it'll surface as an error on whichever LLM node actually tries to call the API. If you're debugging a failed generation and this node looks fine, the config file (not the node) is where to check next.