Textgen Parameters (Venice)

Some nodes generate things; this one generates a parameter bag. Textgen Parameters (Venice) doesn't call the API itself and produces no text you'd ever read - it builds a small dictionary of Venice-specific options and hands it to Generate Text Advanced BETA (Venice) through that node's optional venice_parameters input. If you've ever wanted web search or a canned "character" in your ComfyUI LLM calls, this is how you get it.

What it wires into

Drop it on the canvas, connect its venice_parameters output to the venice_parameters input on the Advanced text node, and you're done. One important gotcha: the plain Generate Text (Venice) node has no venice_parameters input, so this only works with the Advanced node. Hook it to the wrong sibling and nothing will connect.

The inputs

• enable_character + character_slug - Venice has a gallery of public "characters" (personas, basically), each with a slug like the default strawberry-the-cat. The slug is only sent when enable_character is on, so the combo can sit there doing nothing until you flip it. The character list comes from the API, so like the model dropdowns it needs a working key to populate.
• strip_thinking_response - strips <think></think> blocks from the reply. Useful on reasoning models that otherwise decorate their answers with their internal chain-of-thought.
• disable_thinking - on models that support it, turns thinking off entirely (which also strips those blocks). Turn this on when you want fast, direct answers from a reasoning model.
• web_search - auto, on, or off. auto lets the model decide whether to search; on forces it. This is the killer feature if you've been hand-feeding current facts into prompts.
• enable_web_citations - when web search is on, asks the model to cite sources in [REF]0[/REF] format.
• use_venice_system_prompt - whether to layer Venice's own system prompts in alongside yours. Default is off in the shipped node even though the tooltip claims otherwise - one of those small author's-tooltip-vs-reality mismatches you'll find in this pack.

The output

venice_parameters (STRING), which is really a Python dict under the hood. The Advanced node stuffs it straight into the request payload under the venice_parameters key, so what you configure here is exactly what the API receives.

Installing it

Part of the ComfyUI-Venice-API pack - install all seven nodes at once. ComfyUI Manager (search "ComfyUI-Venice-API"), or:

cd ComfyUI/custom_nodes
git clone https://github.com/DraconicDragon/ComfyUI-Venice-API.git
pip install -r ComfyUI-Venice-API/requirements.txt

Restart, then set your venice.ai key in Settings > VeniceAI > API Key and restart again if the character dropdown stays empty - the list is fetched from the API at startup and cached locally. No model downloads, no heavy dependencies, and the whole feature set is still in beta, so expect the author to reshuffle these fields as Venice's API evolves.