echotokens docs

Build with the gateway.

Drop-in OpenAI SDK. 300+ models — chat, vision, image, speech, video, embeddings. Flat USD billing, pass-through pricing, no markups.

Quickstart

Drop in the OpenAI SDK base URL and ship in five minutes.

Streaming

Real-time SSE deltas with the standard OpenAI stream API.

Images

Flux, Nano-Banana, gpt-image — one endpoint, same shape.

Embeddings

Vector embeddings for search, clustering, and RAG.

What is echotokens?

echotokens is an OpenAI-compatible gateway for 300+ AI models — chat, vision, image, speech, video, and embeddings — billed in flat USD with zero markup. Drop-in replace your base_url and your existing code keeps working: prompts, tool calls, streaming, structured outputs, vision inputs. Everything.

python · swap base url
from openai import OpenAI

client = OpenAI(
    base_url="https://api.echotokens.com/v1",
    api_key="sk-echo-...",
)

res = client.chat.completions.create(
    model="claude-opus-4.7",
    messages=[{"role": "user", "content": "Hello, world."}],
)
print(res.choices[0].message.content)
# Response includes `cost_usd_cents` — pure passthrough.

The pricing model

There are no tokens to convert, no per-model multipliers, no surprise quotas. Every response includes a cost_usd_cents field — the exact USD amount we charged your wallet for that single request. That number equals what the upstream provider charged us. We pass it through.

To get started, top up $100 of credit for $15 during the launch window. We bought wholesale; you get the savings.

Where to next?

Pick the surface you need — each guide is a focused 5-minute read with copy-pasteable code:

  • Quickstart — first call in three lines, in Python, Node, or curl.
  • Streaming — Server-Sent Events with the standard OpenAI streaming API.
  • Images — generate with Flux, Nano-Banana, or gpt-image-1.5.
  • Embeddings — vector embeddings for search, clustering, and RAG.
api key first

Before you can call the gateway you need an API key. Head to the keys page in the portal — click "New key", copy the sk-echo-... token, and you're in.

Supported models

A non-exhaustive sample of what you can pass as the model parameter today:

  • Frontier chat: claude-opus-4.7, gpt-5.5-pro, gemini-3-pro-preview
  • Image: flux-2-pro, gpt-image-1.5, nano-banana-2
  • Speech: deepgram-nova-3, elevenlabs-v3
  • Video: veo-3.1-quality, kling-3-master
  • Embeddings: text-embedding-3-large, voyage-3-large

The full list lives in the keys page model picker; new models land within hours of upstream release.