SynapseX Platform API overview

The SynapseX Platform API is an OpenAI-wire-compatible LLM API. If your application already talks to an OpenAI-compatible endpoint, you change two values — the base URL and the API key — and it runs on SynapseX models. When you finish the pages in this section you will have a working API key, a completed chat completion, and streaming output in your own code.

What is the base URL for the SynapseX API?


https://platform.synapsex.ai/api/v1

This is the only supported base URL for the SynapseX LLM API. Every example in this documentation, and the default in the official Python SDK, uses it.

What can you call on it?

Endpoint	Method	What it does
`/chat/completions`	`POST`	Create a chat completion. Supports server-sent-event streaming.
`/models`	`GET`	List the SynapseX model catalog. It is not filtered to what your key may call.

Both endpoints authenticate a workspace API key (sk-synapsex-…) sent as Authorization: Bearer …, and both require a positive prepaid credit balance.

What you need before your first call

Three things, in this order:

A workspace account on platform.synapsex.ai . A personal workspace is created for you on first sign-in.
Credits. A workspace with a balance at or below zero gets an HTTP 402 before any model runs — a valid key on its own is not enough.
An API key, created in the workspace console. The secret is shown exactly once.

The full walkthrough is in Start with the Platform API.

Which request fields are supported?

This is the single most common source of confusion, so it is stated plainly once here and repeated on Chat completions.

The API forwards exactly these fields:

Field	Required	Notes
`model`	Yes	A SynapseX model id.
`messages`	Yes	OpenAI-style `{"role": …, "content": …}` array.
`stream`	No	`true` returns `text/event-stream`.
`temperature`	No	Sampling temperature.
`top_p`	No	Nucleus sampling.
`max_tokens`	No	Cap on generated tokens.

Any other field in the request body is not forwarded and has no effect. Do not rely on tools, response_format, seed, stop, n, logprobs, penalties or user here. Sending them is not an error; they are simply dropped. See OpenAI compatibility for the full boundary list.

How billing works

Pay-as-you-go prepaid credits, where 1 credit = US$1. There is no subscription on the API platform. A zero balance returns HTTP 402 before the model runs, so you cannot overspend. Details are in Credits and spending.

Note that the SynapseX Chat product at chat.synapsex.ai is a separate account with a separate wallet and a different credit unit. Chat credits and API platform credits are not interchangeable.

Verify the API is reachable right now

You can check the error contract without an account. This call is real and returns the exact body shown:


curl -sS https://platform.synapsex.ai/api/v1/models


{"error":{"message":"Missing or invalid API key","type":"authentication_error"}}

That is the SynapseX error envelope. Every error you can receive is documented on API errors.

In this section

Authenticate with a SynapseX API key — create, scope, use and revoke keys.
Chat completions and streaming — the main endpoint, with curl and Python.
List the SynapseX models on the API — read the catalog at runtime, and handle the ids your key cannot call.
Use SynapseX with OpenAI-compatible clients — the stock openai SDK and third-party tools.
API errors and how to handle them — every status code and the correct response.
SynapseX Python SDK — synapsex-platform, typed errors and streaming.