Bing Image Loader

The name is a little misleading - despite "Image" sitting right in the display title, this node actually covers web, image, video, and news search, and it's really a general-purpose "get current information from Bing into my LLM" node rather than an image-specific one. It's part of comfyui_LLM_party's knowledge-base group, sitting alongside the RAG embedding tool as a second way to feed a live agent context it doesn't already have baked in from training.

How it works

You give it search keywords, it queries Bing's search API for the type of result you asked for, and hands back the results as text you can feed into an LLM node - either as extra context for an answer, or into a persona builder if you're compiling a briefing document. Unlike a tool node, this one runs as a direct step in the graph rather than something the LLM decides to call, so it fires every time the workflow executes.

The inputs and outputs that matter

• searchType (web / image / video / news, default web) - this is the field that actually controls what kind of search runs, and it's worth knowing about since the node's name only advertises the image option.
• keywords - your search query.
• paper_num (default 1) - how many results come back.
• bing_api_key - required unless you've already set one in config.ini; the optional is_custom_api and custom_config_id fields let you point this specific node at a different Azure Bing config than your global default, useful if you're running multiple search profiles.
• One output: text - the search results, flattened to a string, ready to wire into an LLM's context.

How to install it

• ComfyUI Manager - search "comfyui_LLM_party", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/heshengtao/comfyui_LLM_party, then restart.

Then pip install -r requirements.txt from inside the comfyui_LLM_party folder, in your ComfyUI Python environment (python_embeded\python.exe -m pip install -r requirements.txt on portable Windows). Per the README, if you want to use Bing (or Google) search tools, you need to enter your bing_api_key (or google_api_key / cse_id) in config.ini - that's an Azure Cognitive Services / Bing Search API key, a separate signup from anything ComfyUI-related, and it's the actual prerequisite here, not a Python dependency.

Common issues & troubleshooting

No results come back. Check bing_api_key first - either the field on the node or the config.ini fallback needs a valid key, and Bing's search API isn't free beyond a limited tier, so an expired trial or exhausted quota will silently return nothing rather than throw an obvious error.

Results are stale-feeling despite this being a "live" search. paper_num defaults to 1 - you're only getting a single result back. Raise it if you want the LLM to have more than one source to work from before it forms an answer.

Confusing this with the embedding tool. bing_loader gets you current web information; advance_ebd_tool builds a searchable index over a document you already have. They solve different problems - grounding in fresh facts versus grounding in a specific file - and mixing them up is a common source of "why isn't my agent finding this" confusion.

The pack isn't loading. Separate from this node, users have hit a real, current install issue where comfyui_LLM_party goes in via Manager but nodes fail to import - usually a CUDA/PyTorch mismatch with no published compatibility list. Rule that out first if bing_loader isn't in your node browser at all.