Name: Zyloo Unified AI API
Brand: Zyloo
Availability: InStock

Step 1

Quickstart

Zyloo speaks the OpenAI API verbatim. If your code talks to api.openai.com, redirect it at our base URL and you're done.

Base URLs

Agentic CLIs — Claude Code, opencode, and the like — point at https://api.zyloo.io. OpenAI-style clients and editors such as Cursor use https://api.zyloo.io/v1.

bash

# 1. Install the official SDK (any language)
npm install openai

# 2. Get your key from https://zyloo.io/dashboard
export ZYLOO_KEY=sk-zy-...

# 3. Make your first call
curl https://api.zyloo.io/v1/chat/completions \
  -H "Authorization: Bearer $ZYLOO_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zyloo/claude-opus-4-7",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

Using an agentic coding CLI? Set its base URL to https://api.zyloo.io and your Zyloo key, then run as usual.

bash

# Claude Code, opencode, and other agentic CLIs
export ANTHROPIC_BASE_URL=https://api.zyloo.io
export ANTHROPIC_API_KEY=$ZYLOO_KEY

# Cursor and other OpenAI-compatible editors
#   Base URL:  https://api.zyloo.io/v1
#   API key:   $ZYLOO_KEY

Integrations

The big five coding CLIs

Every popular terminal agent speaks the OpenAI or Anthropic wire format, so they all run on a Zyloo key — copy the block for your tool and swap in any model from the price index.

Claude Code · Anthropic

Anthropic's terminal agent. Point its Anthropic-format variables at the gateway.

bash

npm install -g @anthropic-ai/claude-code

export ANTHROPIC_BASE_URL=https://api.zyloo.io
export ANTHROPIC_API_KEY=$ZYLOO_KEY
export ANTHROPIC_MODEL=zyloo/claude-opus-4-7-thinking

claude

Codex CLI · OpenAI

OpenAI's open-source CLI. Register Zyloo as a provider in config.toml.

bash

npm install -g @openai/codex
export ZYLOO_KEY=sk-zy-...

toml

# ~/.codex/config.toml
model = "zyloo/gpt-5.5"
model_provider = "zyloo"

[model_providers.zyloo]
name = "Zyloo"
base_url = "https://api.zyloo.io/v1"
env_key = "ZYLOO_KEY"
wire_api = "chat"

opencode · open source

The model-agnostic open-source favourite. Add Zyloo as a custom provider in opencode.json.

bash

npm install -g opencode-ai
export ZYLOO_KEY=sk-zy-...

json

// opencode.json
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "zyloo": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Zyloo",
      "options": {
        "baseURL": "https://api.zyloo.io/v1",
        "apiKey": "{env:ZYLOO_KEY}"
      },
      "models": {
        "zyloo/claude-opus-4-7": {},
        "zyloo/gpt-5.5": {}
      }
    }
  }
}

Aider · open source

The git-native pair programmer. Uses the standard OpenAI-compatible variables.

bash

python -m pip install aider-install && aider-install

export OPENAI_API_BASE=https://api.zyloo.io/v1
export OPENAI_API_KEY=$ZYLOO_KEY

aider --model openai/zyloo/claude-opus-4-7

Qwen Code · Alibaba

Alibaba's coding agent. Reads OPENAI_* variables — nothing else to configure.

bash

npm install -g @qwen-code/qwen-code

export OPENAI_BASE_URL=https://api.zyloo.io/v1
export OPENAI_API_KEY=$ZYLOO_KEY
export OPENAI_MODEL=zyloo/deepseek-v4-pro

qwen

Security

Authentication

Every request must include a bearer token in the Authorization header. Keys are scoped per project and can be revoked instantly from the dashboard.

http

Authorization: Bearer sk-zy-9f3a0e5b...

For local development we recommend storing the key in a .env file and loading it through your runtime — never commit keys to source control.

Catalog

Models

Reference any model by its canonical Zyloo id. Every id is namespaced under zyloo/... so it's unambiguous across providers. The full list of 59 models lives on the Models page.

bash

# A few examples — see /dashboard/models for the full list
zyloo/claude-opus-4-7-thinking
zyloo/claude-opus-4-7
zyloo/gpt-5.5
zyloo/gemini-3.5-flash
zyloo/deepseek-v4-pro
zyloo/grok-4.3

bash

curl https://api.zyloo.io/v1/models \
  -H "Authorization: Bearer $ZYLOO_KEY"

Tip

Models with extended reasoning end in -thinking — for example zyloo/claude-opus-4-7-thinking or zyloo/gpt-5.5-xhigh.

Endpoint

Chat completions

The same JSON shape as OpenAI's /v1/chat/completions. Tools, JSON mode, vision and structured outputs are supported on every compatible model.

import OpenAI from "openai";

const zyloo = new OpenAI({
  apiKey: process.env.ZYLOO_KEY,
  baseURL: "https://api.zyloo.io/v1",
});

const res = await zyloo.chat.completions.create({
  model: "zyloo/gemini-3.5-flash",
  messages: [
    { role: "system", content: "You are concise." },
    { role: "user",   content: "Summarize this PR..." },
  ],
  temperature: 0.2,
  max_tokens: 512,
});

console.log(res.choices[0].message.content);

Endpoint

Image generation

Image models generate pictures from a text prompt at /v1/images/generations — the same request and response shape as OpenAI's Images API. Image models are billed per image, not per token, and appear on the Models page with a per-image price.

const image = await zyloo.images.generate({
  model: "zyloo/your-image-model",   // any model marked "Image" on /models
  prompt: "A watercolor lighthouse at dawn",
  n: 1,                              // 1–4 images per request
  size: "1024x1024",
});

console.log(image.data[0].url ?? image.data[0].b64_json);

bash

curl https://api.zyloo.io/v1/images/generations \
  -H "Authorization: Bearer $ZYLOO_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zyloo/your-image-model",
    "prompt": "A watercolor lighthouse at dawn",
    "n": 1
  }'

How image billing works

Each request is charged price-per-image × number of images delivered. `n` selects how many images to generate (1–4, default 1).
You are only charged for images actually delivered: if the provider fails, times out, or returns an error, the hold is released back to your balance automatically.

Models flagged as supporting edits also accept POST /v1/images/edits with a multipart form body (source image + prompt) at the same per-image price. Models without the flag return a 400 on that endpoint.

bash

curl https://api.zyloo.io/v1/images/edits \
  -H "Authorization: Bearer $ZYLOO_KEY" \
  -F "model=zyloo/your-image-model" \
  -F "image=@photo.png" \
  -F "prompt=Add a red scarf" \
  -F "n=1"

Realtime

Streaming

Pass stream: true to receive Server-Sent Events with the same delta format as OpenAI.

const stream = await zyloo.chat.completions.create({
  model: "zyloo/claude-opus-4-7",
  stream: true,
  messages: [{ role: "user", content: "Tell me a story" }],
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

Reliability

Errors

Zyloo returns OpenAI-compatible error objects. Common codes:

Code	Meaning	Action
401	Invalid key	Rotate from dashboard
402	Insufficient credit	Top up wallet
429	Rate limited	Backoff, we route to a sibling provider
5xx	Upstream failure	Auto-retry with idempotency-key

Ready to ship?

Grab a key and start sending traffic in under a minute.

Get an API key

# 1. Install the official SDK (any language) npm install openai # 2. Get your key from https://zyloo.io/dashboard export ZYLOO_KEY=sk-zy-... # 3. Make your first call curl https://api.zyloo.io/v1/chat/completions \ -H "Authorization: Bearer $ZYLOO_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "zyloo/claude-opus-4-7", "messages": [{"role": "user", "content": "Hello"}] }'

# Claude Code, opencode, and other agentic CLIs export ANTHROPIC_BASE_URL=https://api.zyloo.io export ANTHROPIC_API_KEY=$ZYLOO_KEY # Cursor and other OpenAI-compatible editors # Base URL: https://api.zyloo.io/v1 # API key: $ZYLOO_KEY

// opencode.json { "$schema": "https://opencode.ai/config.json", "provider": { "zyloo": { "npm": "@ai-sdk/openai-compatible", "name": "Zyloo", "options": { "baseURL": "https://api.zyloo.io/v1", "apiKey": "{env:ZYLOO_KEY}" }, "models": { "zyloo/claude-opus-4-7": {}, "zyloo/gpt-5.5": {} } } } }

import OpenAI from "openai"; const zyloo = new OpenAI({ apiKey: process.env.ZYLOO_KEY, baseURL: "https://api.zyloo.io/v1", }); const res = await zyloo.chat.completions.create({ model: "zyloo/gemini-3.5-flash", messages: [ { role: "system", content: "You are concise." }, { role: "user", content: "Summarize this PR..." }, ], temperature: 0.2, max_tokens: 512, }); console.log(res.choices[0].message.content);

const image = await zyloo.images.generate({ model: "zyloo/your-image-model", // any model marked "Image" on /models prompt: "A watercolor lighthouse at dawn", n: 1, // 1–4 images per request size: "1024x1024", }); console.log(image.data[0].url ?? image.data[0].b64_json);

curl https://api.zyloo.io/v1/images/generations \ -H "Authorization: Bearer $ZYLOO_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "zyloo/your-image-model", "prompt": "A watercolor lighthouse at dawn", "n": 1 }'

const stream = await zyloo.chat.completions.create({ model: "zyloo/claude-opus-4-7", stream: true, messages: [{ role: "user", content: "Tell me a story" }], }); for await (const chunk of stream) { process.stdout.write(chunk.choices[0]?.delta?.content ?? ""); }

Code

Meaning

Action

401

Invalid key

Rotate from dashboard

402

Insufficient credit

Top up wallet

429

Rate limited

Backoff, we route to a sibling provider

5xx

Upstream failure

Auto-retry with idempotency-key

Documentation.

Quickstart

The big five coding CLIs

Authentication

Models

Chat completions

Image generation

Streaming

Errors

Ready to ship?

Documentation.

Quickstart

The big five coding CLIs

Authentication

Models

Chat completions

Image generation

Streaming

Errors

Ready to ship?