Connect a Provider
A provider is an AI service that hosts models. Add an API key and every model from that provider becomes available.
Get an API key
Section titled “Get an API key”If you don’t already have one, sign up at your provider’s developer portal:
| Provider | Signup | Capabilities |
|---|---|---|
 Anthropic | console.anthropic.com | Chat |
 OpenAI | platform.openai.com | Chat, Embedding, Image Gen, Video Gen |
 Google Gemini | aistudio.google.com | Chat, Embedding, Image Gen, Video Gen |
 Groq | console.groq.com | Chat |
 Together | api.together.xyz | Chat, Embedding |
 xAI | console.x.ai | Chat, Image Gen, Video Gen |
 OpenRouter | openrouter.ai | Chat, Embedding, Image Gen (multi-provider gateway) |
 Perplexity | perplexity.ai | Chat |
 DeepSeek | platform.deepseek.com | Chat |
 Voyage AI | voyageai.com | Embedding, Reranking |
 Jina | jina.ai | Embedding, Reranking |
 Ollama | ollama.com | Local Chat, Local Embedding |
 LM Studio | lmstudio.ai | Local Chat, Local Embedding |
Ollama and LM Studio don’t need an API key — configure the local server endpoint instead.
Add the key in QARK
Section titled “Add the key in QARK”- Open Settings → Providers
- Select your provider
- Paste the API key
Screenshot: Provider settings with API key field and encryption indicator
Keys are encrypted with AES-256 before they touch disk. No plaintext keys are stored, logged, or sent anywhere except the provider’s API endpoint.
Validate
Section titled “Validate”QARK validates the key automatically after you paste it:
- Green checkmark — valid, connected
- Red warning — invalid, expired, or provider error
If it fails, confirm you copied the full key and that your provider account has API access enabled.
Browse models
Section titled “Browse models”After validation, QARK fetches the provider’s model list (cached for 5 minutes). Models are organized by capability:
| Category | Used for |
|---|---|
| Chat | Conversations, agents, tool use, reasoning |
| Embedding | RAG indexing, semantic search |
| Reranking | Cross-encoder reranking for RAG accuracy |
| Image Gen | Text-to-image in conversations |
| Video Gen | Text-to-video in conversations |
| Local | Models on your hardware via Ollama or LM Studio |
Multiple providers
Section titled “Multiple providers”You can connect as many providers as you want. This lets you compare models in split view, route tasks to cheaper models, or use different providers for chat vs. embedding vs. reranking.
Add or rotate keys anytime in Settings → Providers.