Translate with Any OpenAI-Compatible Model

ButterKit works with any cloud model that exposes an OpenAI-compatible chat completions endpoint. This includes self-hosted models, local inference servers, and third-party providers beyond the ones with dedicated setup guides.

If your provider or tool accepts requests in the same format as the OpenAI /v1/chat/completions API, you can use it for translation in ButterKit.

Prerequisites

• A ButterKit Pro license
• A running endpoint that implements the OpenAI chat completions API format
• An API key (if required by your endpoint)

Connect a Custom Endpoint to ButterKit

Open Settings

In ButterKit, go to Settings > Models.

Add a new model

Click the + button to add a new cloud model.

Enter your model details

Fill in the following fields:

Field	Value
Description	A label for this model (e.g. Ollama Llama or Together AI)
URL	Your endpoint’s base URL (see compatible providers below)
API Key	Your API key, or leave blank for local endpoints that don’t require one
Model	The model identifier your endpoint expects
API Key Header	Leave blank for standard Authorization: Bearer headers. Set to a custom header name if your endpoint requires one.
Temperature	0.35

Temperature controls randomness: lower values (0.0-0.3) produce more literal translations, while higher values (0.4-0.7) allow light paraphrasing. We recommend 0.35 as a good starting point. Some endpoints may clamp or ignore this value.

Security: Your API key is stored securely in the macOS Keychain. It is never sent to ButterKit servers or any third party other than the provider you configure.

Test the connection

Click Test Connection to verify that ButterKit can reach your endpoint. You should see a success confirmation.

Start translating

Toggle Use as default if you want this model for all translations, or select it from the Translate with dropdown when adding a localization.

Compatible Providers

Any service or tool that implements the OpenAI chat completions format will work. Here are some popular options:

Check your provider’s documentation for the exact base URL and supported model names.

When to Use a Custom Endpoint

• Local/offline translation. Run Ollama or LM Studio on your Mac for fully private, offline translation with no API costs.
• Specialized models. Use fine-tuned or domain-specific models not available through major providers.
• Cost optimization. Self-hosted or alternative providers can be significantly cheaper for high-volume work.
• Privacy requirements. Keep all data on your own infrastructure with no external API calls.

Troubleshooting

If Test Connection fails:

• Verify your endpoint is running and reachable (for local servers, check that the process is active)
• Confirm the base URL does not include a trailing /v1/chat/completions path (ButterKit appends this automatically)
• Check that the model name matches exactly what your endpoint expects
• If your endpoint uses a non-standard auth header, set the API Key Header field accordingly