One key. One wallet. One bill.
Don't stitch together 30 SDKs, juggle 30 keys, or reconcile 30 invoices at month-end. infrai puts 14 production modules and the vendors behind them under one contract — called over plain HTTP, no SDK to install — with transparent pricing and per-call metadata. Swap a vendor without touching your code.
- 14
- GA modules
- 392
- API routes
- 0%
- China-AI markup
- 1
- Bill
curl https://api.infrai.cc/v1/chat/completions \-H "Authorization: Bearer $INFRAI_API_KEY" \-d '{"model": "auto", "messages": [{"role": "user", "content": "Explain useEffect in one line"}]}'
One HTTP call to start. No SDK, no install.
Routes across the vendors you already trust
14 modules, one endpoint
Every module is a thin, clean contract over best-in-class vendors. Pick a capability; infrai picks the route.
/v1/emailTransactional email with domain verification, suppression, and delivery tracking.
SMS & OTP
/v1/smsProgrammable SMS, one-time-passcode send and verify, with delivery status.
Scheduling
/v1/schedulingCron jobs, queues, and webhooks — durable background work without a worker fleet.
Observability
/v1/observabilityError capture, events, spans, metrics, and feature flags in one pipe.
Public URL
/v1/public-urlInstant shareable URLs and custom domains for whatever you ship.
Captcha
/v1/captchaHuman-verification widgets and server-side verification across providers.
/v1/pdfGenerate, merge, split, OCR, and watermark documents on demand.
Image Processing
/v1/imageResize, compress, convert, and read metadata through one endpoint.
Realtime
/v1/realtimeChannels, presence, and publish, with auth tokens issued for you.
Storage
/v1/storageBuckets and presigned object access across S3-compatible providers.
Analytics
/v1/analyticsTrack, identify, funnels, and cohorts — product analytics without the wiring.
Billing
/v1/accountBalance, usage, top-ups and invoices — billing without the wiring.
AI Runtime
/v1/aiChat, embeddings, vision, image, speech-to-text and text-to-speech across every major model.
AI Video
/v1/videoText-to-video generation and job tracking across the leading video models.
Account & control plane
Sign-in, wallet, keys, tier, and BYOK — the 74 control-plane routes you never have to wire up yourself.
One contract, every vendor underneath
infrai normalizes the providers below behind stable capability ids — swap vendors without touching your code.
AI models
Video models
SMS
Storage
Realtime
Captcha
Image
Analytics
Payments
Built to stay up — and stay safe
One endpoint in front of every vendor, with failover, idempotency and encrypted keys on by default.
Automatic multi-vendor failover
When a vendor degrades or rate-limits, traffic fails over to a healthy one automatically — cost-capped at 1.5× (up to 3× on Enterprise). Your app keeps calling one stable endpoint.
Idempotent by default
Every write takes an idempotency key, so retries are safe and effects apply exactly once — no double charges, no duplicate sends.
Your keys, encrypted and scoped
BYOK and platform credentials are stored in KMS and shown only once. Scope each key to specific capabilities and lock it to an IP allowlist.
Enterprise-grade compliance
SOC 2 and HIPAA, SSO via SAML/OIDC, full audit logs, data-residency control (Enterprise no-China-route option), and a 99.99% uptime SLA on Enterprise.
Pricing you can actually predict
No minimum markup, no small-request fee. Pick a plan; usage is billed transparently on top.
Standard
- $2 trial credit included
- Wallet up to $500
- BYOK: 8 modules, 30-day trial
- Failover up to 1.5× cost
- Trial credit expires after 30 days; paid top-ups never expire
Pro
or $200/year — save 17%
- Wallet up to $5,000
- 5× rate limits
- BYOK: 8 modules, permanent
- Auto-recharge
- Failover chain
Enterprise
- Invoice post-pay (NET 30/60/90)
- SOC 2 / HIPAA
- SSO / SCIM / audit log
- BYOC / dedicated tenant
- 99.99% SLA · failover up to 3.0× cost
Transparent, usage-based pricing
What you see is what you pay. No minimum markup, no per-request fee — here's exactly how usage is priced.
Chinese AI vendors
0% markup
DeepSeek, Qwen, Hunyuan, Doubao, MiniMax billed at vendor list price — not a cent more.
Western AI vendors
5% markup
OpenAI, Anthropic, Google, Mistral and others — billed at vendor cost plus a flat 5%.
Batch API
100% passthrough
Opt into a 24h SLA and the vendor's 50% batch discount passes straight through to you.
Pricing classes
Call it from anywhere — zero install
No SDK to install. Every capability is a plain HTTPS request to https://api.infrai.cc with a Bearer key — call it with curl, Python, JavaScript, or any language that speaks HTTP.
Works from any language with an HTTP client — Go, Rust, Java, C#/.NET, Ruby, PHP and more.
Zero install. Every language, every editor.
A zero-install REST API and an MCP server — drop infrai into any stack or environment. Every response returns cost, latency and vendor metadata so you always know what each call did.
Zero-install REST API
Plain HTTPS + a Bearer key. Call it from curl, Python, JavaScript, Go, Rust — any language, same capability ids, same metadata, nothing to install.
MCP server
An MCP server exposes infrai’s capabilities to any MCP-compatible environment.
Transparent metadata
cost_usd, latency_ms, vendor, cache_hit, sla_tier on every response.
Every successful call returns:
{
"cost_usd": 0.0021,
"latency_ms": 486,
"vendor": "deepseek",
"cache_hit": true,
"sla_tier": "realtime"
}Available integrations
MCP server
@infrai/mcp-serverClaude Code skill
/infraiCursor rules
.cursorrulesThe standard library for AI-built apps
Sign in once with Google or GitHub to get a key, then call any of 14 unified modules over plain HTTP — no SDK, no install. The backend services your app needs to run, Chinese AI at 0% markup, one wallet, one bill.