vLLM provider

Uses: Kong Gateway AI Gateway Admin API deck KIC Konnect API Terraform

Upstream paths

AI Gateway automatically routes requests to the appropriate vLLM API endpoints. The following table shows the upstream paths used for each capability.

Capability	Upstream path or API
Chat completions	`/v1/chat/completions`

Supported capabilities

The following tables show the AI capabilities supported by vLLM provider when used with the AI Proxy or the AI Proxy Advanced plugin.

Set the plugin’s route_type based on the capability you want to use. See the tables below for supported route types.

Text generation

Support for vLLM basic text generation capabilities including chat, completions, and embeddings:

Capability	Route type	Streaming	Model example	Min version
Chat completions	`llm/v1/chat`	Supported	vllm-llama-3-8b	3.14

vLLM base URL

The base URL is $UPSTREAM_URL, where {route_type_path} is determined by the capability.

AI Gateway uses this URL automatically. You only need to configure a URL if you’re using a self-hosted or vLLM-compatible endpoint, in which case set the upstream_url plugin option.

Configure vLLM with AI Proxy

To use vLLM with AI Gateway, configure the AI Proxy or AI Proxy Advanced plugin.

Here’s a minimal configuration for chat completions:

kong.yaml

Copied!

_format_version: "3.0"
plugins:
  - name: ai-proxy
    config:
      route_type: llm/v1/chat
      model:
        provider: vllm
        name: ai/smollm2
        options:
          upstream_url: ${{ env "DECK_VLLM_UPSTREAM_URL" }}

curl -i -X POST http://localhost:8001/plugins/ \
    --header "Accept: application/json" \
    --header "Content-Type: application/json" \
    --data '
    {
      "name": "ai-proxy",
      "config": {
        "route_type": "llm/v1/chat",
        "model": {
          "provider": "vllm",
          "name": "ai/smollm2",
          "options": {
            "upstream_url": "'$VLLM_UPSTREAM_URL'"
          }
        }
      }
    }
    '

Copied!

Make the following request:

curl -X POST https://{region}.api.konghq.com/v2/control-planes/{controlPlaneId}/core-entities/plugins/ \
    --header "accept: application/json" \
    --header "Content-Type: application/json" \
    --header "Authorization: Bearer $KONNECT_TOKEN" \
    --data '
    {
      "name": "ai-proxy",
      "config": {
        "route_type": "llm/v1/chat",
        "model": {
          "provider": "vllm",
          "name": "ai/smollm2",
          "options": {
            "upstream_url": "'$VLLM_UPSTREAM_URL'"
          }
        }
      }
    }
    '

Copied!

echo "
apiVersion: configuration.konghq.com/v1
kind: KongClusterPlugin
metadata:
  name:
  namespace: kong
  annotations:
    kubernetes.io/ingress.class: kong
  labels:
    global: 'true'
config:
  route_type: llm/v1/chat
  model:
    provider: vllm
    name: ai/smollm2
    options:
      upstream_url: '$VLLM_UPSTREAM_URL'
plugin: ai-proxy
" | kubectl apply -f -

Copied!

resource "konnect_gateway_plugin_ai_proxy" "my_ai_proxy" {
  enabled = true

  config = {
    route_type = "llm/v1/chat"

    model = {
      provider = "vllm"
      name = "ai/smollm2"

      options = {
        upstream_url = var.vllm_upstream_url
      }
    }
  }

  control_plane_id = konnect_gateway_control_plane.my_konnect_cp.id
}

Copied!

This example requires the following variables to be added to your manifest. You can specify values at runtime by setting TF_VAR_name=value.

variable "vllm_upstream_url" {
  type = string
}

Copied!

For more configuration options and examples, see:

AI Proxy examples

AI Proxy Advanced examples

vLLM provider

Upstream paths

Supported capabilities

Text generation

vLLM base URL

Configure vLLM with AI Proxy

Help us make these docs great!

Still need help?