🔱 Artha Gemini Translate

Artha Gemini Translate is the 🔱 Artha pack's translation node: a text block, a from language, a to language, and Gemini does the rest. It exists to solve a genuinely annoying real-world problem - a lot of the best prompts, workflows, and model cards live in languages other than English (Chinese especially), and copy-pasting between Google Translate tabs is clumsy mid-workflow.

It's a simple node, and it stays simple. There's no glossary, no memory, no fine-tuned model to install. You're paying for Gemini's multilingual competence per call, which for natural-language prompt translation is more than good enough.

How it works

The node builds a system prompt telling Gemini to translate, sends your text to the gemini-2.5-flash model (the default) with your chosen language pair, and returns the translation as a plain STRING. That's the whole mechanism - no local models, no API beyond the Gemini generate_content call, and the result is a normal string you can pipe into a text encoder or any other Artha text node.

The language lists are fixed combos, not free text, and they're the languages that actually matter for this community: Chinese, Spanish, English, Hindi, Portuguese, Bengali, Russian, Japanese, Turkish, German, French, Italian. If your pair isn't in the list, this node won't do it - there's no "custom" option.

Inputs that matter

• text - the source text, multiline.
• lang_from - source language combo, defaults to Chinese.
• lang_to - target language combo, defaults to English.
• api_key, model, max_tokens, temperature - the standard Artha Gemini parameter block.

Output is a single response STRING. Wire it into your prompt text input (or a ComfyUI text display) and you're done.

Installing it

It's part of the one-pack Artha install:

cd ComfyUI/custom_nodes
git clone https://github.com/Cyrostar/ComfyUI-Artha-Nodes
pip install -r ComfyUI/custom_nodes/ComfyUI-Artha-Nodes/requirements.txt

then restart ComfyUI - or install "ComfyUI-Artha-Nodes" through ComfyUI Manager and let it handle dependencies. No model files; the only real dependency for this node is google-genai. The catch is the API key: get a free one from aistudio.google.com and put it in the pack's api.json (gemini_api_key), as a GEMINI_API_KEY environment variable, or in the node's api_key field (which the pack warns is visible plain text in your workflow).

Common issues

• Empty response: API key missing or invalid; the node prints the error to the ComfyUI console.
• Language pair not in the dropdown: nothing you can do about it here - the combos are hard-coded. Pick the closest supported pair or use a different translator node.
• It's cloud-only: offline, this node does nothing. It's an API wrapper, not a local translator.
• Token cost: this is a metered Gemini call like every Artha LLM node. For short prompts it's effectively free-tier pennies; for translating huge lore dumps in a batch, watch your quota.

Where does it actually shine? Translating workflow-embedded prompts from Chinese or Japanese sources into English (or the reverse, if you're sharing prompts to non-English communities) without breaking flow. It won't translate your SDXL danbooru tags meaningfully - booru tags aren't language - but for natural-language prompt prose it's exactly the tool.