Frequently Asked Questions
01 Translation Engines
What are the three translation engines?
Apple Translation uses the same
translation built into macOS. It supports 19 languages, is very fast,
and requires no setup beyond downloading the language pair.
Apple Intelligence understands context and nuance better, and supports more languages. You need Apple Intelligence turned on in System Settings.
Built-in AI model is powered by Google's TranslateGemma, a purpose-built translation model running locally via llama.cpp. 55 languages, download once, works forever, fully offline. No Apple features required.
Apple Intelligence understands context and nuance better, and supports more languages. You need Apple Intelligence turned on in System Settings.
Built-in AI model is powered by Google's TranslateGemma, a purpose-built translation model running locally via llama.cpp. 55 languages, download once, works forever, fully offline. No Apple features required.
Which engine should I use?
Start with Apple Translation if
your language is supported. It's the fastest option. If you need a
language it doesn't cover, or want better handling of slang and
nuance, try Apple Intelligence.
If you don't have either of those available, the
built-in AI model works on any Mac.
What sizes does the built-in AI model come in? Which one should I pick?
All three models are from Google's TranslateGemma family,
a purpose-built translation model supporting 55 languages,
running locally via llama.cpp in GGUF format (Q4_K_M quantization):
Small (4B) — ~2.6 GB download. Any Mac with 8 GB RAM.
Medium (12B) — ~7.4 GB download. Needs 16 GB RAM or more.
Large (27B) — ~16.6 GB download. Needs 24 GB RAM or more.
Download any of them with one click from the settings screen. If the in-app download doesn't work, you can download the model files manually from HuggingFace and import them:
TranslateGemma 4B (Small) —
TranslateGemma 12B (Medium) —
TranslateGemma 27B (Large) —
Go to settings, enable AI translation, select "On-Device Model", and click "Import custom model" to load the downloaded file.
Small (4B) — ~2.6 GB download. Any Mac with 8 GB RAM.
Medium (12B) — ~7.4 GB download. Needs 16 GB RAM or more.
Large (27B) — ~16.6 GB download. Needs 24 GB RAM or more.
Download any of them with one click from the settings screen. If the in-app download doesn't work, you can download the model files manually from HuggingFace and import them:
TranslateGemma 4B (Small) —
translategemma-4b-it.Q4_K_M.ggufTranslateGemma 12B (Medium) —
translategemma-12b-it.Q4_K_M.ggufTranslateGemma 27B (Large) —
translategemma-27b-it.Q4_K_M.gguf Go to settings, enable AI translation, select "On-Device Model", and click "Import custom model" to load the downloaded file.
Can I use my own GGUF model?
Yes. Plae supports importing any GGUF model file. Go to settings, enable AI translation, select "On-Device Model" as the backend, and click "Import custom model." The model needs to support the chat/instruct format for best results.
Are there other compatible models I can use?
Yes. Plae ships with TranslateGemma, but any GGUF model that supports the chat/instruct format will work.
Qwen 2.5 Instruct
models are a popular alternative — they're smaller downloads and still produce good translations:
Qwen 2.5 0.5B Instruct —
Qwen 2.5 1.5B Instruct —
Qwen 2.5 3B Instruct —
Download the GGUF file from HuggingFace, then go to settings, enable AI translation, select "On-Device Model", and click "Import custom model" to load it.
Qwen 2.5 0.5B Instruct —
qwen2.5-0.5b-instruct-q4_k_m.gguf (~490 MB)Qwen 2.5 1.5B Instruct —
qwen2.5-1.5b-instruct-q4_k_m.gguf (~1.12 GB)Qwen 2.5 3B Instruct —
qwen2.5-3b-instruct-q4_k_m.gguf (~2.1 GB)
Download the GGUF file from HuggingFace, then go to settings, enable AI translation, select "On-Device Model", and click "Import custom model" to load it.
Where are downloaded models stored?
Models downloaded through Plae are stored in
If you use other llama.cpp-based tools, you may also have compatible GGUF models in
~/Library/Application Support/Plae/Models/.
If you use other llama.cpp-based tools, you may also have compatible GGUF models in
~/Library/Caches/llama.cpp/.
To use those in Plae, click "Import custom model" in settings and select the file.
02 Languages
How many languages does Plae support?
Apple Translation supports 19 languages: Arabic, Chinese (Simplified
and Traditional), Dutch, English (US and UK), French, German, Hindi,
Indonesian, Italian, Japanese, Korean, Polish, Portuguese, Russian,
Spanish, Thai, Turkish, Ukrainian, and Vietnamese. Apple Intelligence
and the built-in AI model can handle additional languages beyond this list.
Do I need to download anything for languages?
For Apple Translation, macOS will prompt you to download the language
pair the first time you use it. It's a small download and only happens
once. For the built-in AI model, you download the model once from the
settings screen. After that, everything works offline.
03 Using the App
How do I open Plae?
Plae lives in your menu bar. Click the icon or press the keyboard
shortcut (default: ⌘⇧T) to open it. You can customize the shortcut in settings.
Does Plae need an internet connection?
No. All three engines run on your Mac. Once you've downloaded the
language pair or AI model, everything works offline. Plae never
connects to the internet.
What is the presentation mode?
Presentation mode shows your translation in large text across your
screen. It's useful when you need to show a translation to someone
sitting across from you, like in a meeting or at a restaurant. Press
⌘⇧L to open it.
What macOS version do I need?
Plae requires macOS Sequoia (15.0) or later.
Apple Translation and the built-in AI model work on any supported Mac.
Apple Intelligence translation requires macOS Tahoe or later.
Does Plae save my translations?
Yes. Plae keeps a local history of your recent translations so you
can go back to them. You can also mark translations as favorites.
Everything is stored locally on your Mac.
04 Pricing
Is Plae a subscription?
No. Plae is a one-time purchase of $4.99 on the Mac App Store. Pay once, use it forever. No recurring fees.
Can I use Plae on multiple Macs?
Yes. Since it's purchased through the Mac App Store, you can download
it on any Mac signed in with the same Apple ID.
How do I get a refund?
Purchases are handled by Apple. Contact Apple Support or use
reportaproblem.apple.com to
request a refund.
05 Privacy
Does Plae send my text anywhere?
No. All translation happens on your Mac. Your text never leaves your
computer. Plae doesn't have a server, a database, or any way to see
what you're translating.
Does Plae collect analytics or telemetry?
No. Zero analytics, zero telemetry, zero tracking. There's no code
in the app that phones home.
How do I contact support?
Email us at support@plae.app.