← Back to all guides

Claude Desktop Custom Endpoint Setup with 59API

Why connect Claude Desktop to a custom endpoint?

If you use Claude Desktop every day, the fastest way to cut model costs is to route requests through a compatible relay instead of paying full direct pricing for every experiment. A custom endpoint setup lets you keep the same desktop workflow while changing where the app sends its API calls. For developers, that means one place to test prompts, compare models, and scale usage without changing habits.

59API is a strong fit for this workflow because it is an AI API relay with cheap, pay-as-you-go access to Claude Opus, Sonnet, Haiku, and Fable, plus GPT models. It uses native, official-quality models with no downgrade, and it is built to work smoothly with Claude Code, Codex, and any OpenAI SDK. The base URL is https://api.59api.com, and the pricing model is designed for people who want to pay only for what they actually use.

Step 1: Create your relay account and key

Start by signing up for 59API and generating an API key from your dashboard. Keep the key handy because Claude Desktop will need it if your build supports a custom provider or endpoint field. If you are running a shared team setup, create a separate key for each person or project so usage stays easy to track. This is especially useful when you want to monitor which prompts or jobs are consuming tokens.

Step 2: Find the endpoint settings in Claude Desktop

Open Claude Desktop and look for the settings area related to developer options, API provider settings, or advanced configuration. Some desktop builds expose a visible custom endpoint field; others require editing a local settings file. The exact layout can vary, but the goal is the same: change the base URL from the default provider to your relay endpoint.

For 59API, enter https://api.59api.com as the base URL. If the app asks for a region, organization, or proxy field, leave it blank unless your environment specifically requires it. When a model selector is available, choose the Claude model you want to use, such as Opus for deep reasoning, Sonnet for balanced work, or Haiku for fast, lightweight tasks.

Step 3: Save, restart, and verify the connection

After you add the key and endpoint, save the settings and fully restart Claude Desktop. A restart matters because desktop clients often cache old connection details until the app reloads. Then open a new chat and run a simple test prompt, such as asking for a short summary, a checklist, or a few lines of code. If the response arrives normally, your custom endpoint is live.

• 401 or unauthorized: double-check the API key and make sure it was copied without extra spaces.
• 404 or not found: confirm the base URL is exactly https://api.59api.com and that the app is not appending an extra path incorrectly.
• Timeouts: try again on a stable network, then lower the request size to isolate whether the issue is prompt length or connectivity.
• Wrong model name: match the model ID expected by your setup, especially if you are switching between Claude and GPT families.

Step 4: Use the setup to control cost

Once the connection works, the real advantage is cost control. A relay like 59API is useful when you are iterating on prompts, testing tools, or running many small experiments where direct premium access adds up quickly. Because it is pay-as-you-go, you are not locked into heavy monthly spend just to keep the workflow available. That makes it a practical choice for solo builders, startups, and teams that want predictable usage-based billing.

Another advantage is flexibility. You can keep Claude Desktop for your day-to-day conversations, then switch models when you need a different tradeoff between speed, depth, and cost. If you also use Claude Code or an OpenAI SDK in the same project, 59API gives you one relay layer instead of juggling separate providers and billing systems.

When a custom endpoint is worth it

Use this setup if you want a lower-cost development loop, if you are testing multiple model families, or if you want the convenience of one endpoint for both Claude and GPT access. It is also helpful when you want to preserve model quality. With 59API, you are not settling for a stripped-down replacement; you are using native-quality models through a cheaper relay.

If you want to try this workflow, sign up for 59API, grab your key, and connect a small test project first. That gives you a low-risk way to validate the endpoint, confirm pricing, and see how much you can save before rolling it out across your full Claude Desktop workflow.

Pricing · Docs