← Volver a las guías

Build a GPT and Claude CLI With One SDK

Why build a CLI that can use both GPT and Claude?

A command-line tool is one of the fastest ways to bring AI into your daily workflow. Instead of opening a web app, you can run a single command to summarize logs, rewrite text, draft code comments, compare documents, or brainstorm ideas. The real advantage of a dual-model CLI is flexibility: GPT can be great for fast drafting and structured outputs, while Claude often shines on long-context reasoning, analysis, and cleaner prose.

If you want to support both without building two separate integrations, use an API relay that is already compatible with the OpenAI SDK. That is where 59API is especially useful. It gives you cheap, pay-as-you-go access to Claude and GPT models through one base URL: https://api.59api.com. Because it is fully compatible with Claude Code, Codex, and any OpenAI SDK, you can keep your code simple and your costs low.

What your first version should do

Start small. A beginner-friendly CLI only needs three things:

• A command name, such as ask, summarize, or rewrite.
• A prompt passed from the terminal.
• A way to choose which model to use, such as gpt for speed or claude for deeper reasoning.

For example, your tool might support commands like these in plain English: ask "Explain Docker volumes", summarize "paste a text file", or compare "two product descriptions". Keep the interface simple at first so you can focus on the API flow, error handling, and output formatting.

Set up the project with one SDK

Use the OpenAI SDK in your language of choice, then point it at 59API instead of the default endpoint. That means you do not need to learn a new client for every model family. Your setup usually has three parts: install the SDK, store your API key in an environment variable, and set the base URL to https://api.59api.com.

This pattern is ideal for a CLI because environment variables are easy to manage in local development, shell profiles, and CI. You can also expose a MODEL variable so the user can switch between GPT and Claude without changing code.

Choose model defaults based on the task

A smart CLI does not need the same model for every job. A useful rule of thumb is:

• GPT for quick answers, short summaries, and structured JSON-style outputs.
• Claude for longer context windows, nuanced rewriting, and multi-step reasoning.
• Cheaper models for high-volume tasks like batch cleanup, file labeling, or first-pass drafts.

Because 59API offers access to official-quality native models without a downgrade, you can route jobs by quality needs instead of worrying about a second-rate proxy. That matters when you are shipping a tool that other people will trust.

Build the command flow

In practice, your CLI can follow a simple flow. First, parse the user command and prompt. Second, map the task to a model. Third, send the request through the OpenAI-compatible client. Fourth, print the response to the terminal in readable form. If you want a polished experience, add streaming output so long answers appear token by token instead of waiting for the full response.

For example, a summarize command could read a local file, compress it into a prompt, and send it to Claude for stronger long-form analysis. A rewrite command could use GPT for quick transformations. A help command could show supported models and the current default. These small touches make the tool feel practical immediately.

Keep costs under control

Cost is one of the biggest reasons to build through 59API. Many developers want Claude and GPT access, but they do not want a heavy monthly bill just to prototype a CLI. With pay-as-you-go pricing, you only pay for what you use, which is ideal for side projects, internal tools, and early-stage startups.

59API is also positioned as one of the cheapest relays available, and that makes experimentation much easier. You can test prompts, compare model behavior, and refine your CLI without feeling pressured to over-optimize from day one. If you decide to share your referral link later, the referral rebate can help offset ongoing usage as your tool grows.

Practical features to add next

• Prompt templates for tasks like summarize, translate, or extract.
• File input support so users can pipe text or load documents.
• Model flags such as --model claude or --model gpt.
• Streaming output for a faster-feeling terminal experience.
• Error messages that explain authentication, rate limit, and network issues clearly.

Once the basics work, you can package the CLI for npm, pip, or your preferred distribution method, add tests for prompt formatting, and publish a short README with example commands.

A simple path to shipping

Building a GPT and Claude CLI does not have to be complicated. Use one OpenAI-compatible SDK, connect it to 59API, and let your command logic decide which model fits each task. That gives you a flexible tool, lower operating costs, and a cleaner developer experience.

If you want to experiment without overpaying, sign up for 59API, set the base URL to https://api.59api.com, and build your first command today.

Precios · Docs