answers.delivery

How-to · 3 min read

How to use GPT, Claude and Gemini with one API key

Developers who want to try GPT, Claude, Gemini and DeepSeek side by side end up with a separate account, key, balance and SDK for every provider. Each one has its own billing page, its own request format and its own way of naming models. Switching your app or coding tool from one model to another turns into a config change in three places. You want one key and one balance, and to change models by changing a single string.

Published
17 Nov 2026
Delivery
No. 12
Topic
AI and agents
Evidence
10 claims quoted from aelder.dev

01
Why it's hard

Every major model vendor ships its own API shape, so code written for one does not run against another without an adapter. Keeping several accounts topped up means several invoices, several spending limits and several dashboards to check when a bill looks wrong. Coding tools make it worse, because each tool only supports some providers natively and expects a specific base URL format. And when you just want to compare two models on the same task, the setup cost often outweighs the experiment. So most people stick with whatever model they set up first, even when another one would do the job better or cheaper.

This part describes the usual way. It makes no claims about AELDER, so nothing is quoted here.

02
How AELDER does it

AELDER gives you one API key for 49 models, including GPT, Claude, Gemini, DeepSeek, Qwen and Kimi, behind an OpenAI-compatible endpoint. Requests can use OpenAI-style Chat Completions, OpenAI-style Responses or Anthropic-style Messages, so existing SDKs work by changing the base URL and key. You pick the model per request with an exact model ID, and every call shows in one dashboard against one balance. Billing is per token from a prepaid balance, with no subscription. Top-ups of $20 or more add 5% to 15% bonus credit, which applies across every model.

Send text requests in OpenAI-compatible Chat Completions, OpenAI-compatible Responses or Anthropic-compatible Messages.Quoted from aelder.dev
Set the model in each request. Same key for all of them, every call in your logs.Quoted from aelder.dev
Any OpenAI-compatible app: set its custom base URL, your key and a model ID.Quoted from aelder.dev

03
Step by step

  1. 1Create an account, top up at least $5 and create a key on the API keys page.
  2. 2Set the base URL to https://api.aelder.dev/v1 for OpenAI SDKs, or https://api.aelder.dev for Anthropic SDKs and Claude Code.
  3. 3Call GET /v1/models with your key and copy the exact model IDs you want.
  4. 4Put the model ID in each request; the key stays the same for every model.
  5. 5Check the dashboard after a test call to confirm the model and token usage.

04
Tips

Test the exact feature you need on the exact model before you rely on it. Tool calls, reasoning output and image input pass through the three formats, but they depend on the model and have not been checked for every one in the catalog.

A model name alone does not guarantee support for every API format or optional parameter.Quoted from aelder.dev

Watch the /v1 suffix. OpenAI SDKs want it in the base URL, while Anthropic SDKs add the path themselves, so the same URL in both places gives you /v1/v1 and a 404.

Check the client's URL convention to avoid /v1/v1.Quoted from aelder.dev

Keep a direct vendor account if you need anything beyond text generation. Embeddings, image generation, audio, files and realtime endpoints are not served, and hosted vendor tools like web search are not covered either.

Hosted tools from the original vendor APIs, such as web search or computer use, are not covered.Quoted from aelder.dev

Store the key in an environment variable or secrets manager, never in browser code or a public repo. One key that reaches every model is convenient, which also means a leaked key can spend your whole balance on any of them.

Keep your key private. Use environment variables in server-side code; do not put it in browser code, public repositories or screenshots.Quoted from aelder.dev

05
Questions

Can I use my existing OpenAI or Anthropic SDK?

Yes. Point the SDK at the AELDER base URL and use your AELDER key; both OpenAI and Anthropic request formats are accepted.

Use the format your client expects, add your key and choose an exact model ID.Quoted from aelder.dev

Does it support embeddings or image generation?

No. Only text requests in the three formats and the model list are served; other endpoints return 404.

Embeddings, image generation, audio, files and realtime endpointsQuoted from aelder.dev

Are model prices fixed?

No. The service is in beta and model prices may change without prior notice.

AELDER is currently in beta. Model prices may change without prior notice.Quoted from aelder.dev

More from AELDER

All 2 answers