Platform capabilities

Everything you need to route, run, and optimize AI.

Purpose-built for teams running AI across multiple models and providers — with routing, data residency, and cost control built in.

Read the docs

One API key, every model

Point your code at Mizan and call OpenAI, Anthropic, Gemini, or models hosted locally in Saudi Arabia — by changing just the model name. Top up your wallet and go.

Spend visibility, everywhere

Track usage across every AI provider — routed through Mizan or connected via admin key — in real time, on one timeline.

Automatic cost optimization

Flags duplicate subscriptions, overspending on pricier models, and gives daily savings advice from your real usage.

Multi-provider support

Access OpenAI, Anthropic, Gemini, Bedrock, locally hosted Saudi models, and more through one integration — no vendor lock-in, switch models instantly.

Projects & scoped API keys

Structure spend into projects, issue scoped API keys per team or app, and track usage globally or per project.

Connect provider admin keys

See spend that never touched Mizan by connecting admin keys from OpenAI, Anthropic, and other providers directly.

Email receipt sync

Sync Gmail or Outlook to automatically backfill and track AI subscription receipts — or upload them manually.

Token intelligence

See exactly which teams, applications, and workflows are driving the highest token consumption.

Cost anomaly detection

Detect unusual spikes in usage instantly — the same day they happen, before costs escalate.

AI budgets & controls

Set spending limits per provider and period, and monitor AI costs against predefined budgets automatically.

Subscription visibility

Monitor every AI subscription and shadow software purchase across teams — no more surprise line items.

Aman — automatic PII redaction

Turn on Aman and every request is scanned to strip PII — emails, credit card numbers, ID numbers, addresses — before it ever reaches an LLM. Built with Saudi data-residency and privacy requirements in mind.

Full-call tracing

Turn on tracing for any Mizan API key to capture every call in detail — tokens, input, output, and cost. Total visibility into your AI usage.

Attribution

Attribute every LLM trace to a project of yours — a PR, a feature, a task, a literature review, you name it. Connect your token spend to your ROI.

Reliability

Never go down. Even when your providers do.

Set fallback models for every request. If your primary provider is slow, rate-limited, or offline, Mizan automatically retries against the next healthy model in your chain — so your app keeps answering.

OpenAI

Anthropic

Gemini

Bedrock

Automatic failover

Mizan

Without fallback routing

One provider goes down or gets rate-limited, and every request that depends on it fails until it recovers.

With Mizan

Define a fallback chain per request. The moment a model is unavailable, Mizan reroutes to the next one you've configured — automatically, with no code changes.

Illustrative — actual resilience depends on your configured fallback chain and each provider's status.

Data residency & zero retention

You choose where it goes. We keep nothing.

A request through Mizan passes through two points: Mizan and the model provider. You decide where each one sits, and by default neither keeps your prompts or outputs.

Your app

Sends the request, gets the answer

RequestAnswer

Data in transit

Mizan

EU · live

KSA · UAE · US · soon

RequestAnswer

Data in transit

Model provider

EUUSKSAUAE

Data residency

Mizan

Fully hosted in the EU today. KSA, UAE and US hosting are coming soon, and you'll be able to select where your data transits.

Provider

Our model list includes EU, US, KSA and UAE based providers. Choose an EU provider and your data stays in the EU.

Zero data retention

Mizan

By default, everything is in transit only. Mizan never keeps your prompts or outputs. We only keep the metrics that power your dashboard.

Provider

All of our providers operate with zero data retention by default.

Opt-in only

Some features need stored content, so they're off unless you turn them on. Enabling them means Mizan stores the related prompts and outputs, currently in the EU, with multi-region coming soon.

Response cachingSouqTracing

Don't enable them and nothing is saved.

Sensitive field? Bring your own key

Connect your own provider account with BYOK for an extra layer of control over who handles your data.

Read the full breakdown

Until KSA, UAE and US hosting is live, requests to non-EU providers still transit Mizan's EU infrastructure.

How it works

Up and running in minutes.

No agents to deploy, no SDKs to install, no code to write.

Top up your wallet

Fund your Mizan Wallet in seconds — no invoices, no waiting on procurement. You're ready to spend the moment you top up.

Generate your Mizan API Key

Create a scoped API key from your dashboard and point your existing code at Mizan. One key, every provider.

Have fun

Access to every model, real-time analytics, and cost reduction techniques — all in one place. Go build.

Route. Track. Reduce.

From tokenmaxxing to efficiency.

Access Mizan

Already running on Mizan

tokens processed: 540M+tokens processed
API calls routed: 26K+API calls routed
reliability, zero failures: 99%+reliability, zero failures