NewPoint any agent at UberLLM — one key, every model, list price or less.
UberLLM

What is an OpenAI-compatible API endpoint?

An OpenAI-compatible API endpoint is one that speaks the exact request and response shapes of the OpenAI API — the same /v1/chat/completions schema, the same streaming format, the same tool-call fields — so the official OpenAI SDK works against it by changing only the base URL and key. UberLLM is OpenAI-compatible (and Anthropic-compatible at /v1/messages), which is why switching to it is a one-line change.

Get an API keySee live prices

The short definition

An endpoint is "OpenAI-compatible" when it accepts the same HTTP requests and returns the same response shapes as the OpenAI API. In practice that means:

  • the POST /v1/chat/completions request body (model, messages, temperature, tools, stream, …) is accepted as-is;
  • the response is the same choices[].message object, with the same usage token counts;
  • streaming uses the same server-sent-events data: chunks ending in [DONE];
  • tool / function calling uses the same tool_calls fields.

Because the contract matches, the official OpenAI SDK — in Python, JavaScript or any language — works against the endpoint when you change only the base URL and the API key. Your application code does not know the difference.

Why it matters

OpenAI-compatibility turned one vendor’s API into a de-facto standard. A compatible endpoint lets you keep the SDK, tooling and framework integrations you already use (LangChain, the Vercel AI SDK, LlamaIndex, dozens of agent frameworks) while changing what sits behind them.

That is the whole migration to UberLLM:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.uberllm.dev/v1",   # the only line that changes
    api_key="ull_your_key_here",
)

resp = client.chat.completions.create(
    model="deepseek/deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello!"}],
    stream=True,
)

See the OpenAI SDK doc for the full set of supported parameters.

Compatible vs. a true drop-in

"Compatible" is a spectrum. Watch for endpoints that accept the request shape but drop streaming, ignore tool calls, or return non-standard usage. UberLLM implements the chat-completions contract including streaming, tool calls and token usage, and additionally exposes an Anthropic-compatible endpoint at /v1/messages so Anthropic SDKs and Claude Code work unchanged too — Anthropic SDK doc.

What UberLLM adds behind the standard endpoint

The endpoint is standard; the routing behind it is not. Every request is sent to the cheapest healthy seller for the model you named, failures before output are retried on the next seller, and the price you see (fee included) is the price you pay. Browse models at /models and live prices at /llm-api-pricing.

Keep reading

FAQ

What does "OpenAI-compatible" mean?

It means the endpoint accepts the same requests and returns the same responses as the OpenAI API — the same chat-completions schema, streaming format and tool-call fields — so the OpenAI SDK works against it by changing only the base URL and key.

Do I need to change my code to use an OpenAI-compatible endpoint?

Only the base_url and api_key. Everything else in your OpenAI SDK code — messages, streaming, tools, usage — stays the same.

Is UberLLM also Anthropic-compatible?

Yes. Alongside the OpenAI-compatible chat-completions endpoint, UberLLM exposes an Anthropic-compatible /v1/messages endpoint, so Anthropic SDKs and Claude Code work unchanged.