Works with Cursor / Claude Code / Cline
Take control of
runaway AI agent costs.
Just change your base_url to add budget controls, auto-stop & failover.
Let your AI agents run overnight — without the bill shock.
Free plan is free forever — no credit card, sign up with Google
Daily limits help curb overnight runaway costs while you sleep
OpenAI / Anthropic / Google AI supportedNo credit card requiredBYOK (use your own API keys)
AI agents are powerful, but without controls they can rack up unexpected costs fast.
API calls per single prompt
Blocks new requests at limit (helps curb unexpected costs)
qzira is not "just a proxy." It's a cost defense system for the AI agent era.
💰
Budget alert notificationsStarter+
Get email alerts when usage hits 50% and 80%. Catch cost spikes before they grow.
🛡
Daily limits + auto-stopPro+
Set daily caps on top of monthly limits. A kill switch that blocks new requests at the limit — helping curb overnight bill shock from runaway agents.
🔄
FailoverStarter+
If OpenAI goes down, auto-switch to Anthropic. Multi-provider redundancy that minimizes downtime risk.
Change one endpoint. Keep your existing code.
Multi-provider unified endpoint
OpenAI, Anthropic, Google — all through one endpoint. Prefix-based routing adapts to new models instantly.
BYOK (Bring Your Own Key)
Use your own API keys. No markup, no middleman margin. Just pay a flat monthly fee for gateway features.
Streaming support
Full SSE (Server-Sent Events) support. ChatGPT-like real-time responses across all providers.
Auto-retry
Automatic retries on transient errors. Exponential backoff to handle API rate limits gracefully.
Global edge network
Powered by Cloudflare Workers. Low-latency from anywhere in the world.
Usage dashboard
Visualize request counts, response times & error rates in real time. Make informed cost-optimization decisions.
Getting Started
Up and running in 3 easy steps
Sign up with Google
Done in 30 seconds. No credit card required.
Get your API key & register provider keys
Generate a qzira API key in the dashboard and register your OpenAI / Anthropic keys.
Change base_url to api.qzira.com/v1
Just update the endpoint in Cursor, Claude Code or Cline settings. No code changes needed.
Setup complete in about 1 minute
Your dashboard after sign-up
See all your API usage, in one place
Request counts, token consumption & model-level logs in real time. Spot unexpected costs before they happen.
Real-time monitoring
Daily request counts & token usage displayed in charts
Request logs
View model, latency & status for every request
CSV export
Download usage data for detailed analysis
Compatibility
Works with your dev tools
OpenAI-compatible endpoint. Just change the URL in your settings.
Cursor
Verified on Pro and above
Claude Code
Messages API supported
Any OpenAI SDK-compatible tool
Just change base_url
Your existing OpenAI SDK code works as-is. Just change the baseURL.
Use your own API keys — no markup. Pay only for gateway features at a fair price.
Why so affordable? qzira uses the BYOK (Bring Your Own Key) model. You pay AI providers directly, so there's zero markup from qzira. The monthly fee covers gateway features only (budget management, auto-stop, failover, etc.).
Your API keys, kept safe
qzira is a BYOK (Bring Your Own Key) service. API key security is our top priority.
Encrypted storage
API keys are encrypted with AES-GCM. Never stored in plain text.
Processed on Cloudflare Workers
Requests are handled at the edge. Keys are decrypted only within the serverless environment — never sent externally.
BYOK — you own your keys
qzira holds no shared keys. You use your own API keys, so you can revoke or rotate them at any time.
Keys never logged
API keys are never included in usage logs. Even in the unlikely event of a log leak, your keys stay safe.
Streaming included in every plan. Upgrade or downgrade anytime.
Free
/mo
1,000 requests/moMonthly API request limit
1 providerNumber of AI providers you can enable simultaneously (OpenAI / Anthropic / Google AI)
1 API keyNumber of gateway API keys you can issue
StreamingReal-time SSE (Server-Sent Events) response output
Retry (1x)Automatic retry on API errors. Recovers from transient failures
Basic dashboardBasic usage analytics: request counts & error rates
Starter
/mo
10,000 requests/moMonthly API request limit
3 providersNumber of AI providers you can enable simultaneously
2 API keysNumber of gateway API keys you can issue
StreamingReal-time SSE response output. All providers supported
Retry (2x) + Failover2 retries on error + auto-switch to another provider on outage
Budget alertsEmail notification when usage hits your budget threshold. Helps manage costs
Pro
/mo
100,000 requests/moMonthly API request limit
Unlimited providersNo limit on enabled providers
5 API keysNumber of gateway API keys you can issue
StreamingReal-time SSE response output. All providers supported
Retry (3x) + Advanced failover3 retries + customizable failover including 429 rate limit handling
Budget alerts + Auto-stopAuto-stop new requests when budget is exceeded. Helps curb overnight cost runaway
Response cacheCache identical request responses to reduce costs
Business
/mo
500,000 requests/moMonthly API request limit
Unlimited providersNo limit on enabled providers
50 API keysNumber of gateway API keys you can issue
StreamingReal-time SSE response output. All providers supported
Retry (3x) + Advanced failover3 retries + customizable failover including 429 rate limit handling
Budget alerts + Auto-stopAuto-stop new requests when budget is exceeded. Helps curb overnight cost runaway
Response cacheCache identical request responses to reduce costs
Semantic cacheAI analyzes similar requests and auto-caches them
Priority supportPriority handling for your inquiries
Scale
/mo
High-volume (custom limits)Custom request limits for large-scale usage
Unlimited providersNo limit on enabled providers
Unlimited API keysNo limit on API keys. Built for large teams
All features unlockedAccess to all current and future features
Retry (5x) + Advanced failoverUp to 5 retries + customizable failover including 429 rate limit handling
Budget alerts + Auto-stopAuto-stop new requests when budget is exceeded. Helps curb overnight cost runaway
Semantic cacheAI analyzes similar requests and auto-caches them
Custom-tailored for reliable, long-term operation
※ AI model usage fees are paid directly to each provider (BYOK model).
※ Limits may be adjusted for fair use.
※ Billed in JPY via Stripe. USD prices are approximate.
※ Coming soon
Unleash your AI agents
with confidence.
Try it now on the Free plan. Setup takes 30 seconds — just sign in with Google.
Start free in 30 secondsNo credit card30-second setupAPI keys encrypted