Generate Text (Venice)

This is the plain-vanilla text node in the ComfyUI-Venice-API pack: send a prompt to venice.ai, get a string back. No local model, no VRAM, just the API key. It does exactly one thing, does it without ceremony, and is honestly the one you'll probably grab when you want a quick answer rather than a tuned generation.

It's also the one the pack author quietly nudges you away from. The README notes the "text gen node is missing but will be updated soon" and points you to Generate Text Advanced BETA (Venice) for the updated experience. The Advanced node is the full-fat version; this one is its little sibling with a fraction of the controls. That's not necessarily a reason to avoid it - sometimes five knobs is five too many - but it is the older, simpler path.

How it works

Same machinery as the Advanced node: it builds a system + user message pair and POSTs to Venice's /chat/completions endpoint, then hands back choices[0].message.content as the response STRING. Nothing runs on your machine. And like its sibling, it's stateless - no chat history, each run is a fresh single exchange.

The inputs

• model - dropdown populated from your Venice account, defaulting to llama-3.3-70b. If it's stuck on the default, the model list never loaded; check your API key and restart.
• system_prompt and prompt - the whole point of the node.
• temperature (default 0.5) and top_p (default 0.9) - your sampling dials.
• frequency_penalty and presence_penalty - both default to 1.5, which is notably more aggressive than the Advanced node's zero. That's an author's-default quirk, not a law; if your outputs read as stiff or evasive, these penalties are probably why. Drop them toward 0 for more natural flow.
• enable_vision + image_for_vision - optional image input for vision-capable models, sent only when the flag is on.

Output

One output: response (STRING). Wire it into a text display node or feed it downstream into another node's prompt field. There's no built-in preview - read it in the outputs panel or display it yourself.

When to reach for which

If you just need a model to write a description, a title, or a one-shot answer, this node is fine and fast to set up. The moment you need max_completion_tokens, repetition_penalty, min_p, dynamic temperature, or the Venice-specific parameters (characters, web search, thinking), switch to the Advanced node - it accepts all of those, and it's where the pack's author puts his energy.

Installing it

Same pack as everything else in ComfyUI-Venice-API - install once, use all seven nodes:

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

Or install via ComfyUI Manager by searching "ComfyUI-Venice-API". Restart, then set your key in Settings > VeniceAI > API Key. Light dependencies, no model downloads - the only "hardware" requirement is a valid Venice account and an API key that can reach /chat/completions.