Your AI stays online,
even during outage

2 lines: to integrate
4 live: providers, 1 more landing

Read the docs

Your base URL

api.metriqual.com/v1

Chat
Voice
Image
Video

One request, mid-outage

openai/gpt-4ostopped answering
anthropic/claude-sonnet-4-5answered

Conversation history carried across. Your user never saw an error.

OpenAI
Anthropic
Gemini
MistralSoon
MiniMax
OpenAI
Anthropic
Gemini
MistralSoon
MiniMax
OpenAI
Anthropic
Gemini
MistralSoon
MiniMax

Why Metriqual

Every model call you make runs through providers you do not control. Metriqual holds the keys, picks the model, replays the request when one stops answering, and prices every call against the key that made it.

Cost

See which model was cheapest, on your own traffic

Every call comes back priced. The dashboard ranks the models your key actually used, so the cheaper one that still answers is a change you can make on purpose.

Rank byCheapestAverage cost per request

1openai/gpt-4o-miniCheapest
2anthropic/claude-sonnet-4-5
3openai/gpt-4o

Product

One key, one URL, every model

Failover that carries the conversation

When a provider stops answering, the request is replayed against the next one in the chain with its history intact.

openai/gpt-4otimeout
anthropic/claude-sonnet-4-5200 OK

Every modality

The same key reaches all of them.

Chat
Voice
Image
Video
Music
Image-to-video

Spend per key

Logged against the key that made the call.

openai/gpt-4o-mini$

anthropic/claude-haiku$

openai/gpt-4o$$

Move in two lines

Change the base URL and the key. The rest of your call stays where it is.

https://api.openai.com/v1

https://api.metriqual.com/v1/chat/completions

-H "Authorization: Bearer mql_your_key"

Providers behind one endpoint

Metriqual holds the provider keys. Your code never sees them, and never changes when the routing does.

OpenAI
Anthropic
Gemini
MistralSoon
MiniMax

Modalities

Chat, voice, and image. Same key.

Streamed chat with tool calls, voice turns with per-stage latency, and image generation, all through one endpoint. Failover covers all three.

Search agents…

support-bot

OpenAI

Anthropic

Knowledge base · 3 docs

What's our refund window, and does it apply to annual plans too?

gpt-4o 640ms212 tokens

Refunds are available within 14 days of purchase. Annual plans are included, prorated for the unused months rather than refunded in full

Ask support-bot something…

Live web search

Setup

How it works

Four steps from your current provider call to a routed one. The migration is the first line of the first step.

Read the docs

Point at one URL
Swap the base URL and the key in the SDK you already use. Nothing else in the call changes.
Metriqual picks the model
It holds the provider keys and routes each call by the rules you set, so your code never carries a provider name.
A failure gets replayed
When a provider stops answering, the request goes to the next one in the chain carrying its conversation history.
Every call comes back priced
Model, key and spend are attached to each request, and a key cannot spend past the ceiling you set.