LLM providers

DlxAI is LLM-neutral. Use cloud credits (default, routed via OpenRouter), or bring your own API keys for any major model.

Three access modes

• Credits (default): Just sign in. The DlxAI backend pays for OpenRouter calls and deducts credits per token.
• Self-hosted: Bring your own API keys. Fully local. Zero cloud dependency.
• Subscription / coding plans: Connect Claude / Gemini / Kimi etc. subscription accounts. No API key needed.

Supported providers

Mainstream

• OpenAI (GPT-4 / 4o / o1)
• Anthropic (Claude Opus / Sonnet / Haiku)
• Google (Gemini Pro / Flash)
• xAI (Grok)

Open source / cost-effective

• DeepSeek
• Mistral
• Groq (ultra-fast inference)
• OpenRouter (aggregates everything)

Chinese providers

• Zhipu AI / Z.ai (GLM family)
• Moonshot / Kimi
• Qwen (Alibaba)
• Volcengine / Doubao (ByteDance)
• MiniMax
• Xiaomi / MiMo

Local

• Ollama — run open-source models on your own machine, zero cloud

Enterprise / cloud-native

• Amazon Bedrock
• NVIDIA NIM
• Venice AI

Add an API key

1. Click Providers in the sidebar
2. Pick a provider and click "Add key"
3. Paste the key (encrypted via Keychain / DPAPI on save)
4. Pick a default model
5. Save — applies immediately

Per-provider proxy

In a restricted region? Configure HTTP / SOCKS5 proxies per provider or per API key. The DlxAI proxy router routes traffic by hostname automatically, leaving other traffic untouched.