SGLang Provider: Self-Hosted OpenAI-Compatible Server

SGLang exposes open-weight models through an HTTP API that is compatible with OpenAI. The openai-completions provider family in OpenClaw handles connections to SGLang and automatically finds the models that are available.

Property	Value
Provider id	`sglang`
Plugin	bundled, `enabledByDefault: true`
Auth env var	`SGLANG_API_KEY` (any non-empty value if server has no auth)
Onboarding flag	`--auth-choice sglang`
API	OpenAI-compatible (`openai-completions`)
Default base URL	`http://127.0.0.1:30000/v1`
Default model placeholder	`sglang/Qwen/Qwen3-8B`
Streaming usage	Yes (`supportsStreamingUsage: true`)
Pricing	Marked external-free (`modelPricing.external: false`)

When you enable SGLANG_API_KEY, OpenClaw will auto-discover the models that SGLang serves. To keep discovery dynamic while also specifying a custom base URL for SGLang, set sglang/* inside agents.defaults.models. Refer to Model discovery (implicit provider) below.

Getting started

Start SGLang

Start SGLang with an OpenAI compatible server. Your base URL must expose /v1 endpoints, for instance /v1/models or /v1/chat/completions. SGLang is typically deployed on:

http://127.0.0.1:30000/v1

Set an API key

If your server has no authentication configured, any value is accepted:

export SGLANG_API_KEY="sglang-local"

Run onboarding or set a model directly

openclaw onboard

Alternatively, define the model manually:

{
  agents: {
    defaults: {
      model: { primary: "sglang/your-model-id" },
    },
  },
}

Model discovery (implicit provider)

When SGLANG_API_KEY is configured (or an auth profile is present) and models.providers.sglang is not set, OpenClaw sends requests to:

GET http://127.0.0.1:30000/v1/models

and turns the returned IDs into model entries.

Note

If models.providers.sglang is set explicitly, OpenClaw defaults to using the models you declared. To have OpenClaw also query the /models endpoint of that configured provider and incorporate every advertised SGLang model, add "sglang/*": {} to agents.defaults.models.

Explicit configuration (manual models)

Use explicit configuration when:

SGLang is running on a different host or port.
You need to fix contextWindow or maxTokens values.
Your server demands a real API key, or you want to control the headers.

{
  models: {
    providers: {
      sglang: {
        baseUrl: "http://127.0.0.1:30000/v1",
        apiKey: "${SGLANG_API_KEY}",
        api: "openai-completions",
        models: [
          {
            id: "your-model-id",
            name: "Local SGLang Model",
            reasoning: false,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 128000,
            maxTokens: 8192,
          },
        ],
      },
    },
  },
}

Advanced configuration

Proxy-style behavior

SGLang functions as a proxy style OpenAI compatible /v1 backend, not as a native OpenAI endpoint.

Behavior	SGLang
OpenAI-only request shaping	Not applied
`service_tier`, Responses `store`, prompt-cache hints	Not sent
Reasoning-compat payload shaping	Not applied
Hidden attribution headers (`originator`, `version`, `User-Agent`)	Not injected on custom SGLang base URLs

Troubleshooting

Server not reachable

Make sure the server is active and answering requests:

curl http://127.0.0.1:30000/v1/models

Auth errors

If authentication errors occur, supply a valid SGLANG_API_KEY that matches your server's setup, or explicitly configure the provider under models.providers.sglang.

Tip

When SGLang is running without authentication, any non-empty SGLANG_API_KEY value is enough to enable model discovery.

Model selection, How to pick providers, model references, and failover behaviour.
Configuration reference, The full config schema, including provider entries.