LLMs
One URL, one format, every provider. Change the model field and nothing else.
Status: 🟢 Live, running on the gateway today
OpenAI
Support: GPT (OpenAI)
support-gpt
Support: GPT mini (OpenAI)
support-gpt-mini
Support: FAQ (small, cached)
support-faq
Anthropic
Support: Claude (Anthropic)
support-claude
Support: Claude Sonnet (Anthropic)
support-sonnet
Support: Claude Haiku (Anthropic)
support-haiku
Ollama (local open weights)
Support: open weights (local)
support-oss
Support: Qwen (local)
support-qwen
Support: Gemma (local)
support-gemma
Multi provider failover
All models
| Model alias | Provider | Upstream model | Guardrails | |
|---|---|---|---|---|
| support-gpt | OpenAI | gpt-5.6-terra | 6 policies |
| support-claude | Anthropic | claude-opus-5 | 5 policies |
 | support-gemini | gemini-3.8-flash | 5 policies | |
| support-oss | Ollama (local open weights) | llama3.2:3b | 5 policies |
| support-sonnet | Anthropic | claude-sonnet-5 | 5 policies |
| support-haiku | Anthropic | claude-haiku-5-5 | 5 policies |
| support-gpt-mini | OpenAI | gpt-5.4-mini | 6 policies |
| support-gemini-pro | gemini-3.1-pro-preview | 5 policies | |
| support-qwen | Ollama (local open weights) | qwen3.8:27b | 5 policies |
| support-gemma | Ollama (local open weights) | gemma4:latest | 5 policies |
| | support-auto | Multi provider failover | gpt-5.6-terra then claude-opus-5 then gemini-3.8-flash | 6 policies |
| support-faq | OpenAI | gpt-5.4-nano | 3 policies |
Call any model
curl http://localhost:8010/llm/chat/completions \
-H "apikey: $KONGNET_API_KEY" -H "Content-Type: application/json" \
-d '{"model": "support-gpt", "messages": [{"role": "user", "content": "Why is my wifi slow?"}]}'
The gateway URL is the demo data plane. See the LLM Gateway API for the full reference.