Guides

Migration

unoblox is a drop-in for the OpenAI and Anthropic APIs. In most codebases the switch is two lines — the base URL and the key — because the request and response schemas are the ones your SDK already speaks.

From OpenAI

The gateway implements the OpenAI Chat Completions API natively at /v1/chat/completions. Point the OpenAI SDK at the gateway, use your ub-gw-… key, and change model names to author/model slugs. Fields you already send — messages, tools, tool_choice, response_format, stream — are forwarded to the model untouched.

python

import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.unoblox.ai/v1",   # changed
    api_key=os.environ["UNOBLOX_API_KEY"],       # changed (ub-gw-…)
)

# Everything else is identical. Model ids become author/model slugs —
# browse /models for a real one.
resp = client.chat.completions.create(
    model="<model-id>",                          # was e.g. "gpt-4o"
    messages=[{"role": "user", "content": "Hello"}],
    # tools, tool_choice, response_format, stream, temperature, …
    # all pass straight through to the model.
)
print(resp.choices[0].message.content)

node.js

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.unoblox.ai/v1", // changed
  apiKey: process.env.UNOBLOX_API_KEY,      // changed (ub-gw-…)
});

const resp = await client.chat.completions.create({
  model: "<model-id>",                      // was e.g. "gpt-4o"
  messages: [{ role: "user", content: "Hello" }],
});

From Anthropic

The gateway also exposes the Anthropic Messages API at /v1/messages, so you can keep your Anthropic-shaped payloads. One important difference:

unoblox accepts both Anthropic authentication forms.

The gateway accepts both Authorization: Bearer <key> and x-api-key: <key> on /v1/messages. With the Anthropic SDK, use auth_token for Bearer authentication or api_key for the default x-api-key header.

curl · /v1/messages

curl https://api.unoblox.ai/v1/messages \
  -H "Authorization: Bearer $UNOBLOX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<model-id>",
    "max_tokens": 256,
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

anthropic sdk

import os
from anthropic import Anthropic

# unoblox accepts either Authorization: Bearer or x-api-key.
# Use auth_token for Bearer authentication, or api_key for x-api-key.
# base_url omits /v1: the SDK appends /v1/messages itself.
client = Anthropic(
    base_url="https://api.unoblox.ai",
    auth_token=os.environ["UNOBLOX_API_KEY"],
)

msg = client.messages.create(
    model="<model-id>",
    max_tokens=256,   # required by the Messages API
    messages=[{"role": "user", "content": "Hello"}],
)
print(msg.content[0].text)

Prefer to consolidate on one client? Anthropic users can also just use the OpenAI SDK against /v1/chat/completions (shown above) and select an anthropic/… model — the gateway translates for you.

What changes, what doesn't

Aspect	Before	With unoblox
Base URL	api.openai.com / api.anthropic.com	api.unoblox.ai/v1
Key	sk-… / x-api-key	ub-gw-… as Authorization: Bearer
Model id	gpt-4o, claude-…	an author/model slug from /models
messages, tools, response_format, stream	as you send them	mostly forwarded — see each skin's own reference page for what it validates or rejects
Billing	a separate foreign-currency invoice per provider	one prepaid ₹ balance, metered per token

The gateway validates model and a non-empty messages array first; most other fields are then forwarded to the model, but each skin also rejects a handful of fields it does not support (unsupported tool types, certain Responses/Anthropic-only fields, the :online/:batch suffixes) rather than passing them through silently — see API reference.

Frameworks

Anything built on the OpenAI protocol works the same way. For LangChain, construct ChatOpenAI with the gateway base URL and your key:

langchain

import os
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.unoblox.ai/v1",
    api_key=os.environ["UNOBLOX_API_KEY"],
    model="<model-id>",
)

LlamaIndex, the Vercel AI SDK, LiteLLM and similar tools follow the same pattern — set the OpenAI base URL to https://api.unoblox.ai/v1 and the key to your ub-gw-… token.

Migrate incrementally.

unoblox extensions — multi-model models: [] failover, unoblox/auto routing, BYOK — are all optional additions to the same request. Move traffic over first, adopt them when you want them.

PreviousAnalytics API