Claude Desktop
Route Claude Desktop through the Zuplo AI Gateway. The gateway authenticates, meters, and routes every Chat, Cowork, and Code session started from the app.
To give Claude Desktop tools from a Zuplo MCP route instead, see Connect Claude Desktop and Claude.ai.
Prerequisites
-
Create an Anthropic provider in the AI Gateway.
-
Create an app for Claude Desktop and assign it to the team.
-
Copy the API URL and API Key from the app page.
Configure Claude Desktop
-
Open Developer → Configure Third-Party Inference.
-
Under Connection, set Inference provider to Gateway.
-
Under Gateway credentials, set Gateway base URL to the app's API URL, for example
https://my-gateway-main-2e18f50.zuplo.app/config_fe0a04972d2848e0a94ae4b8bcd1497e, and Gateway API key to the app's API key. -
Leave Credential kind set to Static API key and Gateway auth scheme set to Bearer.
-
Under Models, add each model the app may use. See Add models.
-
Click Apply locally. Claude Desktop saves the configuration and relaunches.
Rules
- Gateway base URL is the app's API URL without
/v1. Claude Desktop appends/v1/messages. - Gateway auth scheme stays Bearer. The gateway reads the API key from
the
Authorization: Bearerheader. - Credential kind stays Static API key. AI Gateway apps authenticate with API keys, not identity-provider tokens.
Add models
Replace the provider name
The examples assume your provider is named anthropic. Replace that prefix with
the name of your provider before entering any model. A provider named
my-anthropic routes my-anthropic/claude-sonnet-5; a Bedrock provider named
bedrock routes bedrock/anthropic.claude-sonnet-5. The model part must be an
ID that provider serves. Check your provider's page
in the Portal.
The Model list replaces the models Claude Desktop discovers on its own. Click Add once per model:
- Model ID — the full
providerName/modelreference, for exampleanthropic/claude-sonnet-5. A bare ID fails with a400. - Display name — the name in the model picker. Leave it blank and Claude Desktop formats one from the ID.
- Offer 1M-context variant — off, unless your provider accepts 1M-token context for the model.
- Tier alias — the Claude tier (Opus, Sonnet, Haiku, or Fable) this model
stands in for. The bare alias, for example
opus, then resolves to it.
The first entry is the picker default. Include a Haiku-tier model: Claude Desktop runs background and sub-agent tasks on a small, fast model.
The app's Model Filtering policy controls which models the app can use.
Verify
Send a message in Chat and start a Cowork session. Both appear in the app's usage in the Zuplo Portal, as do Code sessions started from the desktop app. Terminal Claude Code sessions use their own configuration. See Claude Code.
Troubleshooting
| Error | Fix |
|---|---|
400 The request body model must use "providerName/model" | A model list entry has no provider prefix. Give every entry a full providerName/model ID. |
Deploy to a fleet
Use the configuration window's Export menu instead of Apply locally. It
produces a .mobileconfig profile for macOS MDM tools such as Jamf, a .reg
policy file for Intune or Group Policy, and related artifacts. Managed
configuration overrides local settings. See Anthropic's
Deploy Claude Desktop with an LLM gateway
and Deploy with MDM.