AI Providers & Native Keychain Proxy

Connect an account and choose it in compatible AI actions.

Core WorkflowsCaptured interfaces · 0:08 loop

Captured client interfaces with sample data; blue outlines are guide annotations.

Key Highlights

20 Built-in Provider Profiles

Covers OpenAI / Codex, OpenAI Codex (OAuth), OpenRouter, DeepSeek, Doubao, Google / Gemini, Anthropic, xAI, xAI Login (OAuth), z.ai / GLM, Kimi (Global & CN), MiniMax (Global & CN), Alibaba Cloud, Tencent TokenHub, Xiaomi MiMo, Hugging Face, LM Studio, and Custom Endpoint.

Host-managed credentials and native requests

When a compatible extension uses a SwiftBiu provider account, the native proxy injects that account’s credential into requests without handing it to the extension.

Browser OAuth with Auto Refresh

Sign into OpenAI Codex or xAI Login in your browser with PKCE loopback callback and automatic background token refresh.

Chat Completions vs Responses API Preview

Toggle API Format between /chat/completions and /responses with a live "SwiftBiu will call: POST ..." endpoint preview.

Multimodal Model Capability Manager

Sync remote model lists, inspect Image/Video/Audio input badges, add custom models, and recategorize models across Chat, Image, Video, Voice, and Advanced.

Smart Extension Form Binding

Compatible AI extensions automatically hide manual key inputs and present live Account + Model dropdowns.

Custom Configuration

View all 8 custom options & thresholds
Setting / Control What it controls
Add Provider Adds a new API-key or OAuth AI provider account to macOS Keychain.
Provider Platform Selects from 20 built-in cloud/local AI platforms with prefilled endpoints and auth modes.
API Key Stores your secret key in Keychain (optionally E2EE-synced via iCloud Keychain when Secure API Keys sync is enabled).
Custom API Base URL Overrides the default base URL for self-hosted gateways, proxies, or local LM Studio/Ollama servers.
API Format Switches compatible endpoints between Chat Completions (/chat/completions) and Responses API (/responses) with live POST URL preview.
Open Console Jumps directly to the provider’s official API key or billing portal.
Models Opens the cached model catalog grouped into Chat, Image, Video, Voice, and Advanced with Sync Models, Add Custom Model, and Move to… category tags.
Details Displays Token Preview, Fingerprint, Length, Refresh Token status, Scope, Expiry, and guarded Copy Token / Reveal Token actions.

Settings Path & Prerequisites

Settings menu:AI Providers
  • An API key from any supported platform, a local LM Studio server, or a ChatGPT / xAI account for OAuth login

Troubleshooting

Expand common questions & fixes (3)

Can third-party extensions steal my API key?

Host-managed provider credentials are injected by the native proxy. A key entered directly into an extension’s own settings is handled according to that extension’s configuration and implementation.

My custom gateway uses /v1/responses instead of /chat/completions?

Edit the provider account, switch API Format to Responses API (/responses), and verify the live POST URL preview before saving.

Why didn’t my OAuth login sync to my second Mac?

For security, OAuth tokens stay bound to the local Mac’s Keychain; sign in once on each Mac, whereas API keys can sync via iCloud Keychain.