One API Key for Claude, GPT-5,
Gemini & DeepSeek
Standard OpenAI SDK drop-in replacement. Switch between 17+ frontier models, leverage automated provider failover, and pay with non-expiring pay-as-you-go credits.
OpenAI SDK Compatible
Drop-in base URL change
Auto-Failover Routing
Zero rate limits & 99.9% uptime
Pay-As-You-Go Credits
Never expire, start with $5
No Monthly Lock-In
Save $60+/mo on subscriptions
Interactive Developer Documentation & SDK HubNew
Explore full endpoints reference, Claude Code, Cursor, Python & Node.js SDK guides, and recipes.
One Unified API Instead of 4 Expensive Subscriptions
Stop maintaining separate API keys and monthly minimums for OpenAI, Anthropic, Google, and DeepSeek.
Universal Drop-In SDK
Zero learning curve. Keep using the official openai library in Python, Node.js, Vercel AI SDK, and LangChain. Just change the base URL.
Automated Zero-Downtime Failover
Never suffer from Anthropic 529 outages or OpenAI rate limits. Our smart gateway automatically load balances and fails over to redundant enterprise cloud replicas.
Pay-As-You-Go Credits
Instead of paying $60+/month across separate subscriptions, buy non-expiring credit passes starting at $5. Get 20 free test credits instantly upon sign up.
Ready to start building in under 60 seconds?
Sign in to claim your 20 free credits and generate your API key immediately.
How the Unlocked AI Gateway Works
Connect your application in 3 simple steps without refactoring your codebase.
Generate Your Secret Key
Sign in with Google or Email. Create a secret API key (sk-unlocked-...) with one click. No credit card required.
Configure Base URL
Point any standard OpenAI client or IDE (Cursor, VS Code, LangChain) to https://unlockedai.net/v1.
Stream All 17 Models
Pass any model parameter (claude-sonnet-5.5, gpt-5, o3, gemini-4-argon) and stream responses instantly.
Public SDK Quickstarts
Copy-paste integration code for your preferred programming language or AI tool.
from openai import OpenAI
# 1. Initialize standard OpenAI client pointing to Unlocked AI Gateway
client = OpenAI(
base_url="https://unlockedai.net/v1",
api_key="sk-unlocked-YOUR_API_KEY"
)
# 2. Call ANY model: Claude Opus, GPT-5, Gemini 4 Argon, DeepSeek R1
response = client.chat.completions.create(
model="claude-sonnet-5.5", # or "gpt-5", "gemini-4-argon", "o3", "deepseek-r1"
messages=[
{"role": "system", "content": "You are a senior systems engineer."},
{"role": "user", "content": "Explain microservices load balancing in 2 sentences."}
],
temperature=0.7
)
print(response.choices[0].message.content)Full API Endpoints & Reference Guide
Detailed parameter schemas, authentication instructions, and response formats for every gateway endpoint.
/api/v1/chat/completionsExecute chat, reasoning, multimodal analysis, and code generation across any of the 17+ frontier models (Claude Opus/Sonnet 5.5, GPT-5, o3, Gemini 4 Argon, DeepSeek R1). Compatible with any OpenAI SDK.
Required Headers
Request Parameters (JSON Body)
| Parameter | Type | Required | Description |
|---|---|---|---|
| model | string | Required* | Model identifier (e.g. claude-sonnet-5.5, gpt-5, o3) or smart alias (auto, frontier, fast). |
| models | string[] | Optional | Fallback routing array: e.g. ["claude-sonnet-5.5", "gpt-4o"]. Automatically fails over if primary model hits upstream outages. |
| messages | object[] | Required | List of message objects: [{"role": "system"|"user"|"assistant", "content": "..."}]. |
| stream | boolean | Optional | Defaults to false. When true, tokens stream over Server-Sent Events (SSE). |
| temperature | number | Optional | Sampling temperature (0.0 to 2.0). Higher values like 0.8 are more creative; lower values like 0.2 are more deterministic. |
| response_format | object | Optional | Structured outputs: e.g. {"type": "json_object"} enforces valid JSON responses. |
| tools | object[] | Optional | Standard OpenAI tool definitions for function calling and AI agent execution. |
Test Queries Directly in Your Browser
Logged-in developers can stream test prompts directly to Claude 3.7 Sonnet, GPT-5, o3, and Gemini 4 Argon with real-time latency meters (ms), token metrics, and instant credit tracking.
Transparent Cost & Token Calculator
Calculate your exact project spend without subscriptions or surprise end-of-month cloud bills.
Credits never expire. No monthly renewal fees. If you don't call the API for a month, you pay $0.
Supported Models & Parameter IDs
Pass these exact strings into the model parameter of your API request.
| Model | Model Parameter ID | Tier | Cost |
|---|---|---|---|
| Claude Opus 5.5Anthropic | frontier | 8 credits/ prompt | |
| Claude Sonnet 5.5Anthropic | frontier | 5 credits/ prompt | |
| GPT-5OpenAI | frontier | 5 credits/ prompt | |
| GPT-4oOpenAI | frontier | 5 credits/ prompt | |
| OpenAI o3OpenAI | reasoning | 5 credits/ prompt | |
| Gemini 4 ArgonGoogle | frontier | 5 credits/ prompt | |
| Gemini 3.1 ProGoogle | frontier | 5 credits/ prompt | |
| Gemini 2.5 ProGoogle | frontier | 5 credits/ prompt | |
| DeepSeek R1DeepSeek | reasoning | 2 credits/ prompt | |
| Gemini 3.8 FlashGoogle | fast | 1 credit/ prompt | |
| Gemini 3.5 Flash-LiteGoogle | fast | 1 credit/ prompt | |
| Gemini 2.5 FlashGoogle | fast | 1 credit/ prompt | |
| DeepSeek V3DeepSeek | fast | 1 credit/ prompt | |
| GPT-5 MiniOpenAI | fast | 1 credit/ prompt | |
| GPT-4o MiniOpenAI | fast | 1 credit/ prompt | |
| Llama 3.3 70BMeta | fast | 1 credit/ prompt | |
| Claude Haiku 5.5Anthropic | fast | 1 credit/ prompt |
Automated Multi-Provider Load Balancing
Our gateway manages throughput, latency, and automated failover transparently across redundant global cloud infrastructure.
Zero-Downtime Failover
If an official model endpoint experiences an outage or latency spike, requests automatically fall back to redundant enterprise cloud replicas.
Sub-Millisecond Token Streaming
Standard Server-Sent Events (SSE) stream tokens with real-time throughput directly to your application or user interface.
Budget & Margin Protection
API keys draw directly from your shared non-expiring credit pool. You never face surprise end-of-month cloud bills.
Start Building with 17+ Models Today
Join thousands of developers using Unlocked AI for high-speed, cost-effective frontier AI inference.