0 models live

Every frontier model.One base URL.

Claude, GPT, Gemini, Grok, DeepSeek, Kimi and GLM through OpenAI, Anthropic and Gemini compatible APIs. Pay per token at published prices. ORYNX never stores your prompts.

Browse models Read docs

Base URL

https://api.orynx.dev/v1

Works with the tools you already use

Each one links to a tested setup guide.

One key for four API formats

Call any model through OpenAI Chat, OpenAI Responses, Anthropic Messages or the native Gemini API, with the SDK you already use. The key and base URL stay the same, and streaming works in all four.

OpenAI Chat: /v1/chat/completions
OpenAI Responses: /v1/responses
Anthropic Messages: /v1/messages
Gemini: /v1beta/models/{model}:generateContent

Connecting

POST/v1/chat/completionskimi-k3

Request

curl https://api.orynx.dev/v1/chat/completions \-H "Authorization: Bearer $ORYNX_API_KEY" \-H "Content-Type: application/json" \-d '{"model": "kimi-k3","messages": [{ "role": "user", "content": "What does an API gateway do?" }]}'

curl https://api.orynx.dev/v1/chat/completions \-H "Authorization: Bearer $ORYNX_API_KEY" \-H "Content-Type: application/json" \-d '{"model": "kimi-k3","messages": [{ "role": "user", "content": "What does an API gateway do?" }]}'

curl https://api.orynx.dev/v1/responses \-H "Authorization: Bearer $ORYNX_API_KEY" \-H "Content-Type: application/json" \-d '{ "model": "gpt-5.5", "input": "What does an API gateway do?" }'

curl https://api.orynx.dev/v1/messages \-H "x-api-key: $ORYNX_API_KEY" \-H "anthropic-version: 2023-06-01" \-H "Content-Type: application/json" \-d '{"model": "claude-sonnet-5", "max_tokens": 512,"messages": [{ "role": "user", "content": "What does an API gateway do?" }]}'

curl "https://api.orynx.dev/v1beta/models/gemini-3.8-flash:generateContent" \-H "x-goog-api-key: $ORYNX_API_KEY" \-H "Content-Type: application/json" \-d '{ "contents": [{ "parts": [{ "text": "What does an API gateway do?" }] }] }'

Response

Waiting for the first token…

{"model": "kimi-k3","choices": [{ "message": { "content": "It gives your services one front door that checks keys and forwards each request." }, "finish_reason": "stop" }],"usage": { "prompt_tokens": 14, "completion_tokens": 16 }}

{"model": "kimi-k3","choices": [{ "message": { "content": "It gives your services one front door that checks keys and forwards each request." }, "finish_reason": null }],"usage": { "prompt_tokens": 14, "completion_tokens": 16 }}

{"model": "gpt-5.5","status": "completed","output": [{ "type": "output_text", "text": "It checks the caller's key, then sends the request to the right backend." }],"usage": { "input_tokens": 14, "output_tokens": 15 }}

{"model": "gpt-5.5","status": "in_progress","output": [{ "type": "output_text", "text": "It checks the caller's key, then sends the request to the right backend." }],"usage": { "input_tokens": 14, "output_tokens": 15 }}

{"model": "claude-sonnet-5","content": [{ "type": "text", "text": "A single entry point that checks who is calling and forwards each request." }],"stop_reason": "end_turn","usage": { "input_tokens": 15, "output_tokens": 15 }}

{"model": "claude-sonnet-5","content": [{ "type": "text", "text": "A single entry point that checks who is calling and forwards each request." }],"stop_reason": null,"usage": { "input_tokens": 15, "output_tokens": 15 }}

{"modelVersion": "gemini-3.8-flash","candidates": [{ "content": { "parts": [{ "text": "It puts one URL in front of many services and routes each call to the right one." }] }, "finishReason": "STOP" }],"usageMetadata": { "promptTokenCount": 9, "candidatesTokenCount": 18 }}

{"modelVersion": "gemini-3.8-flash","candidates": [{ "content": { "parts": [{ "text": "It puts one URL in front of many services and routes each call to the right one." }] }, "finishReason": null }],"usageMetadata": { "promptTokenCount": 9, "candidatesTokenCount": 18 }}

{
  "model": "kimi-k3",
  "choices": [{ "message": { "content": "It gives your services one front door that checks keys and forwards each request." }, "finish_reason": "stop" }],
  "usage": { "prompt_tokens": 14, "completion_tokens": 16 }
}

0 ms14 tokens

Models: 0
Families: 0
API formats: 4; Chat, Responses, Messages, Gemini
Families available: 0/0; Uptime follows the first checks.

What ORYNX does

It forwards each request to the model you choose and bills it per token at the published price.

One API for every family

Call Claude, GPT, Gemini, Grok, DeepSeek, Kimi and GLM with the SDKs you already use. Switch models by changing the model id.

	OpenAI Chat	OpenAI Responses	Anthropic Messages	Gemini

Your prompts are not stored

ORYNX keeps request metadata for billing: model, token counts, cost, timing. Prompts and completions pass through and are never written to disk or logs.

Stored: model, tokens, cost, latency, status
Never stored: prompt, completion, files, images

Published per-token prices

Every price is published per 1M tokens, including cache reads and writes and any long-context or peak-hour tiers. You pay exactly the listed rate.

Check your key anytime

Key lookup shows remaining credit, expiry, token totals and recent requests. No account or password needed.

Long streams stay open

If a model pauses to reason, ORYNX sends a heartbeat after 15 seconds of silence, so proxies and clients don't drop the connection.

Three steps to your first request

1
Get a key
One key with its own credit, expiry and models.
2
Set the base URL
Point your SDK or app at ORYNX.
3
Build
Send requests and track spend in Key lookup.

main.py

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["ORYNX_API_KEY"],
    base_url="https://api.orynx.dev/v1",
)

reply = client.chat.completions.create(
    model="claude-sonnet-5",
    messages=[{"role": "user", "content": "Hello"}],
)
print(reply.choices[0].message.content)

Common questions

Is ORYNX compatible with the OpenAI and Anthropic SDKs?

Yes. Use the official OpenAI, Anthropic and Google GenAI SDKs and change only the base URL and API key. Chat Completions, Responses, Messages and Gemini generateContent are all supported, with streaming.

Do you store my prompts or responses?

No. ORYNX stores request metadata only (model, token counts, cost, timing and status) so it can bill you and show usage. The upstream model providers process your requests and may keep data under their own policies; see the privacy policy.

How is a request priced?

Input, output and cached tokens are multiplied by the model's published price per 1M tokens. Some models charge more for very long prompts or at peak hours; those tiers are listed on every model page.

What happens when my credit runs out or my key expires?

Requests are refused with an error in your client's API format. A key's validity is counted in days from the day it was issued. Contact the person or team who issued your key to top it up or extend it.

Can I use it with Claude Code, Codex or Cursor?

Yes. The docs have tested configuration for Claude Code, Codex CLI, Cursor, OpenCode, OpenClaw, Hermes Agent, Cline, Kilo Code, Continue, Cherry Studio and more.

How do I check my balance?

Open Key lookup and paste your key. Apps that read OpenAI-style billing endpoints, such as NextChat, can also show your credit and expiry.

Pick a model and send your first request

Browse live prices or follow a guide for your favourite tool.