Skip to content

Quick Start

yaku works out of the box with the hosted backend — no API key needed. If you prefer to use your own LLM API key, see Option B below.

Option A: Use the hosted service (default)

Section titled “Option A: Use the hosted service (default)”
  1. Translate something

    With no configuration, yaku sends requests to the hosted service at api.yakulang.com:

    Terminal window
    yaku --to en "Bonjour le monde"

    Output:

    Hello, world
  2. Set a default target language

    Skip typing --to every time:

    Terminal window
    yaku config set default-target en

    Now you can simply run:

    Terminal window
    echo "Bonjour le monde" | yaku
  3. Translate a file

    Terminal window
    yaku -f README.fr.md -o README.md

    yaku detects the .md extension and preserves Markdown structure (headings, code blocks, links) automatically.

The hosted service uses refined, per-language prompts for higher-quality translations. See Hosted Service & Plans for quota limits and how to upgrade.

  1. Set your API key

    yaku supports Gemini, OpenAI, and Anthropic. Pick one:

    Get a free API key from Google AI Studio.

    Terminal window
    yaku config set backend gemini
    yaku config set api-key YOUR_GEMINI_API_KEY

    Or set it via environment variable:

    Terminal window
    export YAKU_BACKEND=gemini
    export GOOGLE_API_KEY=YOUR_GEMINI_API_KEY
  2. Set a default target language

    Terminal window
    yaku config set default-target en
  3. Translate something

    Terminal window
    echo "La réunion est prévue pour demain à 10 heures." | yaku

    Output:

    The meeting is scheduled for tomorrow at 10 o'clock.
Terminal window
# Translate text directly
yaku --to en "おはようございます"
# Translate from a specific language
yaku --from fr --to en "L'API a renvoyé une erreur 404."
# Translate a JSON i18n file, preserving keys
yaku --to en -f zh-TW.json -o en.json
# Polish your writing in another language
yaku --mode polish --to es "Quiero informarte que el deployment será delay."
# Add domain context for better terminology
yaku --to en --context "cooking recipe" -f recipe.fr.md
# Use a project glossary
yaku --to en --glossary terms.yaml -f docs.zh-TW.md
# Stream output in real time
yaku --to en --stream -f article.fr.md
# See model and token usage
yaku --verbose --to en "こんにちは"