GuidesFeatured
Cherry Studio Guide: Your AI Workstation | Part 1
One desktop workspace for conversations, documents and AI tools. Start this four-part Cherry Studio with Omnirouter series with a practical first workflow.
omnirouter4 min read
62 models across 13 vendors
Every AI model.One simple API.
Build with Claude, GPT, Gemini, Grok, DeepSeek, and more through a single reliable endpoint. Pay only for what you use, with no subscription and no provider juggling.
No credit cardOpenAI compatibleNo monthly minimum
api.omnirouter.li / requestsAll systems operational
Incoming request
req_8a21fPOST/v1/chat/completionsclaude-sonnet
Your app
Smart router
Provider A
Selected · 98ms
Provider B
Ready · 121ms
Provider C
Ready · 146ms
Response streamed1,284 tokens842ms$0.0031
FastStreaming responses
AutomaticProvider failover
Per requestTransparent billing
Built for production
The model layer your team does not have to maintain
A focused API surface for shipping, monitoring, and paying for AI workloads without stitching together vendor accounts.
One stable interface
Switch between leading models without replacing your SDK or rebuilding the integration.
Resilient by default
Requests route through ranked providers and fail over automatically when one slows down.
Every cent explained
See tokens, latency, model, and the exact charge for every request in one clear ledger.
Keys stay private
API keys are stored as secure hashes. Provider credentials never enter your application.
Prepaid, not locked in
Add credit when you need it. No subscriptions, seat limits, contracts, or expiring balance.
Providers handled for you
We manage supplier accounts, routing rules, and provider changes behind stable model names.
Change two lines. Keep everything else.
Point your existing OpenAI client at omnirouter and use the model identifier you want. Streaming, tools, and familiar errors work as expected.
quickstart.py
from openai import OpenAI
client = OpenAI(
base_url="https://omnirouter.li/v1",
api_key="sk_live_...",
)
response = client.chat.completions.create(
model="claude-haiku-4-5",
messages=[{"role": "user", "content": "Hello"}],
) 200 OK · 842ms · $0.0031
Know before you send
Simple pricing you can calculate
No tiers to decode. Add prepaid credit and each request is charged at its model's published rate. The estimate uses the same catalog as billing.
Unlimited API keysBalance never expires
See how billing works Estimate your monthly spend
Adjust usage to see a live estimate.
Estimated monthly spend
$105.00
Anthropic list price
$350.00
You save
$245.00
70% less
No subscription. You are charged per request, from prepaid credit. List price: openrouter.
Live catalog
Start with the right model
A few of the most affordable chat models available now.
| Model | Input / 1M | Output / 1M |
|---|---|---|
GPT 6 Lunagpt-6-luna | $0.01 | $0.05 |
DeepSeek V4.1 Flashdeepseek-v4-1-flash | $0.03 | $0.06 |
GPT 5.6 Lunagpt-5-6-luna | $0.012 | $0.072 |
DeepSeek V4 Flash 0731deepseek-v4-flash-0731 | $0.014 | $0.075 |
Qwen3.5 Flashqwen3-5-flash | $0.20 | $0.20 |
From the blog
How the gateway actually behaves
Routing, pricing and failover, written up in detail.
ComparisonsFeatured
The 2026 AI Price War: What Cheaper Frontier Models Change
On 22 September 2026 Anthropic and OpenAI both cut prices. Mozilla measures the open-weight gap at 4.4 months. What that means for your token bill.
The omnirouter team7 min read
Comparisons
Open-Weight AI Models: Choose a Budget Default
Choose a default model with a small, repeatable evaluation. Compare open-weight candidates on correctness, useful output, latency and actual spend.
omnirouter4 min read
Your next model is one API call away
Create an account, add credit when you are ready, and make your first request in minutes.
Questions before you start? Message us on Telegram at @Omnirouter_support