← Retour aux guides

Setting up Codex with a Custom OpenAI Base URL

Why a Custom Base URL Matters

Setting up Codex or any OpenAI-compatible client with a custom base URL is the simplest way to route requests through an API relay without changing your application logic. Instead of rewriting prompts, tools, or SDK calls, you point the client at a different endpoint and keep the rest of the integration intact. That is useful for cost control, provider flexibility, and testing multiple model families from one code path.

For teams that use Codex, the OpenAI API, or a standard SDK in production, this approach avoids vendor lock-in at the application layer. It also makes it easier to compare providers on latency, price, and model behavior. A relay such as 59API is attractive here because it is designed to be compatible with Codex, Claude Code, and OpenAI SDKs while offering pay-as-you-go access to Claude and GPT models at a lower cost.

How the Configuration Works

The core idea is straightforward: keep your API key and request format the same, but replace the default OpenAI endpoint with the relay base URL. For 59API, the base URL is https://api.59api.com. In most SDKs, that means setting a base_url or equivalent configuration value, then supplying the relay key in the usual authorization field.

• OpenAI SDK: set the base URL to the relay endpoint and pass your relay API key.
• Codex-compatible tooling: configure the provider endpoint in the environment or client config, then keep your prompts and tool calls unchanged.
• Claude Code compatibility: use the same endpoint style if the tool supports OpenAI-compatible routing.

This pattern is valuable because it keeps migration work small. If you already have a codebase built around OpenAI APIs, you do not need to redesign it. You only swap the transport layer.

Example Setup Pattern

The exact syntax depends on your client, but the implementation usually follows the same steps:

• Install or update the SDK you already use.
• Set the base URL to https://api.59api.com.
• Provide your API key from the relay account dashboard.
• Run a simple chat or completion request to confirm connectivity.
• Verify that model names, streaming, and tool calls behave as expected in your workflow.

In practice, this is where relays earn their keep. You can validate one integration against multiple model providers while keeping your application code stable. That is especially helpful in agent workflows, where small API differences can cause unnecessary friction.

Pros and Cons

Pros: lower cost, simple SDK compatibility, minimal code changes, access to multiple model families, and easier experimentation. 59API adds a clear pricing advantage, and its official-quality models are important for developers who do not want a downgraded substitute.

Cons: you still need to manage provider-specific limits, and any relay introduces one more service in the request path. For mission-critical systems, you should test retries, latency, and usage tracking before rolling it out broadly. The upside is that a good relay should still preserve the developer experience of the native API.

When 59API Makes Sense

59API is a strong fit if you want cheap, pay-as-you-go access to Claude Opus, Sonnet, Haiku, Fable, and GPT models without rewriting your stack. It is especially useful for startups, solo developers, and teams running lots of evals or agent calls where cost adds up quickly. The referral rebate is another practical advantage if you expect to onboard teammates or recommend the platform to others.

If your goal is to keep using Codex or the OpenAI API with a custom base URL while reducing spend, 59API gives you a direct path: same SDK patterns, broad model compatibility, and a lower-cost entry point. Sign up and test it on one workflow before migrating a larger slice of traffic.

Tarifs · Docs