Skip to main content

Integrations

A published model service is a standard OpenAI-compatible endpoint, so any client or framework that speaks that API can call it by setting its base URL and API key — no Lens-specific client required.

curl​

curl http://<edge-host>:<edge-port>/v1/chat/completions \
  -H "Authorization: Bearer lens-mk-..." \
  -H "Content-Type: application/json" \
  -d '{"model":"Qwen/Qwen3-8B","messages":[{"role":"user","content":"Hello"}]}'

OpenAI Python SDK​

from openai import OpenAI

client = OpenAI(
    base_url="http://<edge-host>:<edge-port>/v1",
    api_key="lens-mk-...",
)
response = client.chat.completions.create(
    model="Qwen/Qwen3-8B",
    messages=[{"role": "user", "content": "Hello"}],
)

LangChain / LlamaIndex / other OpenAI-compatible frameworks​

Any framework that accepts a custom base_url and api_key for an "OpenAI-compatible" provider works the same way — point it at http://<edge-host>:<edge-port>/v1 with a Lens API token (lens-mk-..., issued from the Model services page).

Lens Assistant's own provider​

Lens Assistant and agentic deployment planning go the other direction: Lens is the client, calling out to an external provider you configure (OpenAI, Azure OpenAI, Anthropic, or a self-hosted OpenAI-compatible server — which can itself be another Lens model service).