One endpoint. Every model. Lower spend.
EzRouter is the simplest way to route LLMs across providers, offering curated models at unbeatable prices.
Routed across suppliers
Anthropic
OpenAI
Google
Moonshot
Zhipu
DeepSeek
Platform
Why EzRouter
One API, curated models
Every model worth routing, behind a single OpenAI-compatible endpoint. We only add suppliers that earn their place.
Browse the leaderboard
Dynamic routing and failover
Requests route to the most stable, cheapest supplier every time. When one wobbles, traffic silently moves — no client change needed.
How routing works
Pay as you go
No subscriptions. No per-user pricing. No minimums. Every model is priced below its official API — and the effective rate drops automatically as your usage grows.
Get an API keyModels
The leaderboard
| # | Model | Provider | Intelligence | Price · in / out · per 1M tokens | Context |
|---|---|---|---|---|---|
| 01 | GPT-5.5 | OpenAI |
100
|
$5/$30 | 1M |
| 02 | Claude Opus 4.7 | Anthropic |
99
|
$5/$25 | 1M |
| 03 | Gemini 3.1 Pro Preview |
95
|
$4/$18 | 1M | |
| 04 | GPT-5.4 | OpenAI |
94
|
$2.5/$15 | 1M |
| 05 | Gemini 3.5 Flash |
92
|
$1.5/$9 | 1M | |
| 06 | Kimi K2.6 | Moonshot / Kimi |
90
|
$1/$4 | 256K |
| 07 | Claude Sonnet 4.6 | Anthropic |
86
|
$3/$15 | 1M |
| 08 | DeepSeek V4 Pro | DeepSeek |
86
|
$0.44/$0.87 | 1M |
| 09 | GLM-5.1 | Z.ai |
85
|
$1.4/$4.4 | — |
Illustrative snapshot. Live prices and health scores sync in the dashboard; the full catalog lives in the docs.
Calculator
What it costs, in your numbers
EzRouter
$—
you save —
official API
$—
— more expensive
OpenRouter
$—
— more expensive
Support
Common questions
How do I get an API key?
Sign up and a key is issued in the dashboard immediately. You can create multiple
keys, scope them to projects, and revoke any one without affecting the others.
How is pricing calculated?
Per input and output token, priced below each provider's official API. Our routing
layer picks the cheapest stable supplier for every request — you see the effective rate in your usage logs.
Which models are supported?
A curated set across Anthropic, OpenAI, Google, Moonshot, Zhipu and a few others. We
don't chase every model — we keep the ones that earn their place. The full list lives in the Models catalog.
Is my request data used for training?
No. EzRouter passes requests through to the provider you're routed to and returns the
response. We don't retain prompt or completion payloads beyond short-lived debugging logs you control.
What happens if a provider goes down?
Our router watches latency, error rate and capacity across every supplier in real
time. If one wobbles, the next request silently goes to the next-best provider for that model — no client
change needed.
How do I migrate from OpenAI or OpenRouter?
Swap the base URL in your existing client to the EzRouter endpoint. The request and
response shapes match the OpenAI chat-completions schema, so existing SDKs and tooling keep working.