Build with Adorbis

What are you building?

Choose a path to jump to the relevant integration pattern. Compatibility can vary by client, model, and feature.

Make your first API call

Follow these steps in order. You only need an Adorbis API key, a model ID, and the endpoint URL. The examples below call the Chat Completions interface.

curl https://api.adorbistech.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_ADORBIS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "adorbis/quick",
    "messages": [{"role": "user", "content": "Hello from Adorbis"}]
  }'

# Install the dependency first: python -m pip install requests
import os
import requests

response = requests.post(
    "https://api.adorbistech.com/v1/chat/completions",
    headers={
        "Authorization": f"Bearer {os.environ['ADORBIS_API_KEY']}",
        "Content-Type": "application/json",
    },
    json={"model": "adorbis/quick", "messages": [
        {"role": "user", "content": "Hello from Adorbis"}
    ]},
    timeout=60,
)
response.raise_for_status()
print(response.json())

const apiKey = process.env.ADORBIS_API_KEY;
const response = await fetch(
  "https://api.adorbistech.com/v1/chat/completions",
  {
    method: "POST",
    headers: {
      "Authorization": `Bearer ${apiKey}`,
      "Content-Type": "application/json"
    },
    body: JSON.stringify({
      model: "adorbis/quick",
      messages: [{ role: "user", content: "Hello from Adorbis" }]
    })
  }
);
if (!response.ok) throw new Error(`Request failed: ${response.status}`);
console.log(await response.json());

$headers = @{ Authorization = "Bearer YOUR_ADORBIS_API_KEY" }
$body = @{
  model = "adorbis/quick"
  messages = @(@{ role = "user"; content = "Hello from Adorbis" })
} | ConvertTo-Json -Depth 5
Invoke-RestMethod -Method Post `
  -Uri "https://api.adorbistech.com/v1/chat/completions" `
  -Headers $headers `
  -ContentType "application/json" `
  -Body $body

Authentication

Send your Adorbis API key in the Authorization header. JSON request bodies use the JSON content type.

Protect your key

• Keep keys in server-side environment variables or a secret manager.
• Do not commit keys to source control or expose them in browser JavaScript.
• Rotate a key if you believe it has been exposed.
• Do not log authorization headers.

Model interfaces

Use a documented identifier in the request's model field. Identifiers do not guarantee behavior, availability, or output quality.

API endpoints and full URLs

Choose the endpoint that matches the request format your application sends. Copy the full URL into your client configuration or HTTP request. Full HTTP endpoint URLs below use https://api.adorbistech.com as the server root. OpenAI-compatible clients that ask for a versioned base URL commonly use https://api.adorbistech.com/v1; use the field format requested by the client.

Common request headers

Chat Completions

Use POST https://api.adorbistech.com/v1/chat/completions for OpenAI-compatible chat requests.

{
  "model": "adorbis/quick",
  "messages": [{"role": "user", "content": "Explain strict typing briefly."}],
  "stream": false
}

Streaming

The route supports real Server-Sent Events (SSE) streaming. With stream: true, clients receive incremental chunks when the active execution path supports streaming.

Messages API

Use POST https://api.adorbistech.com/v1/messages for the Anthropic-style Messages compatibility surface. Send a JSON request with a supported model and message content; authenticate with a Bearer API key.

POST /v1/messages
Authorization: Bearer YOUR_ADORBIS_API_KEY
Content-Type: application/json

{
  "model": "adorbis/quick",
  "messages": [{"role": "user", "content": "Hello"}],
  "max_tokens": 256
}

This is an illustrative compatibility pattern, not a promise that every native Anthropic field is supported. Verify model, parameters, tool behavior, and response schema against your intended use before production.

Streaming may emit a buffered/synthetic event lifecycle after the upstream response has been generated. Do not assume token-by-token generation.

Responses API and Codex

Codex connects through Adorbis's dedicated Responses compatibility endpoint at /v1/responses.

The compatibility layer accepts Responses-style requests and translates supported inputs and function-call messages. Do not assume every hosted tool or every Responses feature is supported.

Responses streaming

The current implementation can emit a buffered/synthetic Responses event lifecycle. The upstream response may be generated in full before the event sequence is emitted. Do not assume native token-by-token streaming.

Claude compatibility

Adorbis provides a dedicated /claude/v1/* compatibility surface for Claude-compatible integrations.

Known routes include GET /claude/v1/models, POST /claude/v1/messages, and POST /claude/v1/chat/completions. Feature behavior differs by route; do not assume complete Anthropic API parity.

Claude Messages streaming

/claude/v1/messages supports a buffered/synthetic Anthropic-compatible event lifecycle; the upstream response may be generated first.

Claude chat completions streaming

/claude/v1/chat/completions streaming is not currently verified as supported. Classification remains pending deeper implementation verification.

Streaming matrix

Native incremental upstream chunks and API-compatible event sequences emitted after generation are not equivalent.

Tool and function calling

Adorbis can transport supported tool/function definitions and model-generated tool-call messages. Your application executes the tool. Adorbis does not automatically execute arbitrary customer tools.

1. Your application sends a request and supported tool definitions.
2. The model may return a tool/function call if supported.
3. Your application validates permissions and executes the tool.
4. Your application sends the tool result back in the expected format.
5. The model may then produce a final response.

Embeddings

Adorbis exposes POST https://api.adorbistech.com/v1/embeddings. Authenticate using a Bearer API key and send a JSON request compatible with the embedding interface enabled for your account.

Embedding-specific RPM, TPM, credit, and concurrency behavior must be verified separately; do not infer it from chat-completions limits.

Credits and usage

Credit exhaustion or a configured spending limit can result in HTTP 402 Payment Required. Do not assume a monthly refill schedule, non-expiration policy, or account balance unless current account terms explicitly establish it.

Rate limits

Rate limits can depend on account entitlement, route, and current configuration. This public guide intentionally does not publish tier-specific RPM or daily credit caps until the live configuration and customer-facing plan terms are reconciled.

Handling HTTP 429

The reviewed Adorbis-generated rate-limit response does not include a Retry-After header. Use bounded exponential backoff with jitter, stop after a configured retry budget, and distinguish Adorbis-generated limits from upstream provider errors. Confirm current behavior against the route you use.

Errors and HTTP statuses

Exact response bodies vary by route and failure source. Do not assume every upstream provider error is returned like an Adorbis-generated error.

MCP

A remote MCP surface has been referenced, but this page intentionally does not publish a copy-ready connection URL. The production host, transport, authentication exchange, capability list, and client configuration have not been verified together, so a certified setup recipe cannot be provided yet.

Before connecting a client

1. Obtain the current official MCP connection details from Adorbis.
2. Confirm the transport and authentication method supported by that configuration.
3. Review the exposed tools and their permissions before enabling them.
4. Test connection failures, authorization boundaries, and tool behavior in a non-production context.

BYOK-compatible applications

Applications that support a custom API base URL and corresponding request schema may be configured to use Adorbis. Compatibility is not universal; validate the expected protocol, parameters, streaming behavior, and tool-call handling.

If the client asks for a base URL, use the base URL field. If it asks for a full endpoint URL, use the full endpoint. Do not append /v1 twice. Enter your API key in the client's secret/API-key field, not in the URL.

Compatibility guide

Exact UI labels and configuration fields can change between client versions. Confirm the expected protocol before configuring a client.

Model migration

Adorbis can reduce provider-specific integration coupling by giving an application a stable documented interface. A model change is not necessarily zero-effort.

• Test representative prompts and edge cases.
• Validate response formats and structured output.
• Check parameter and tool-call behavior.
• Measure latency and usage on your own workload.
• Keep a rollback path for application changes.

Model ID guidance

Use only model identifiers published in the current Adorbis account or documentation. A model ID shown in an example may not be enabled for every account or use case; verify access before deploying an integration.

What remains your responsibility

Adorbis provides documented API and compatibility interfaces. You remain responsible for output validation, application logic, tool permissions and execution, secret storage, security, privacy, compliance, error handling, retries, and testing before production changes.