AI Providers
Supported Providers
SoAgents ships with 15+ AI model providers ready to use:
Anthropic
- Anthropic (Subscription) — For Claude Pro/Max subscribers, sign in via browser
- Anthropic (API) — API key auth, supports Claude Sonnet 4.6, Opus 4.6, Haiku 4.5
International Providers
- Google Gemini — Gemini 2.5 Pro, 2.5 Flash, 3.1 Pro Preview
- OpenRouter — GPT-5.4, Claude, Gemini and more
- DeepSeek — DeepSeek Chat, DeepSeek Reasoner
China-based Providers
- Moonshot AI (Kimi) — Kimi K2.5, K2 Thinking, K2
- Zhipu AI — GLM 4.7, GLM 5, GLM 4.5 Air
- MiniMax — M2.5, M2.5 Lightning
- VolcEngine — Aggregates Doubao, GLM, DeepSeek, Kimi models
- SiliconFlow — Multi-model aggregation
- Aliyun Bailian — Qwen 3.5+, Kimi K2.5, GLM 5
- ZenMux — Multi-model aggregation
Setup
- Open Settings → Providers
- Select a provider and enter your API Key
- Save — the model is now available in chat
Advanced Options
Each provider supports:
- Custom Base URL — For self-hosted proxies or private deployments
- Model Alias Mapping — Map sonnet/opus/haiku to custom models
- Timeout Settings — Custom request timeout
- Proxy Settings — HTTP/SOCKS5 proxy support
Switching Models
Switch models at any time:
- Change the default provider in Settings
- Different conversations can use different models
- Token stats update automatically per model
Custom Providers
Add your own provider if the built-in list doesn’t cover your needs:
- Settings → Providers → Add Custom
- Enter name, Base URL, and API Key
- Select API protocol (Anthropic Messages / OpenAI Chat Completions)
- Configure the available model list