One endpoint, every model
Point any OpenAI-compatible client at one base URL. Change the model field to switch between Claude, Gemini, and dozens more — same key, same code.
Public API access
New Claude Fable 5 — Anthropic's most intelligent model — is now live on the gatewayLaunch apps with one prepaid AI API wallet. Users buy credit, generate scoped keys, and route text, code, image, video, and automation calls through an OpenAI-compatible endpoint.
0% service fee until 31 Dec 2026 No subscription · OpenAI-compatible · you pay only the model's rate.
API base
Point any OpenAI-compatible client at the SuperbAPI base URL and pass your key. Account, billing, and support live in the workspace — your code only ever talks to one route.
https://www.superbapi.com/v1
How it works
Wire up SuperbAPI once, then switch models, scope keys, and track spend without touching your code again.
Point any OpenAI-compatible client at one base URL. Change the model field to switch between Claude, Gemini, and dozens more — same key, same code.
Generate a key per app or customer. Set a monthly budget, an expiry date, and an allowed-model list on each — edit them any time.
Top up once. Every call draws from a single balance at the model's exact rate — with tokens and spend visible per key. No subscription, and 0% service fee until 31 Dec 2026.
What you can call
Route text, code, image, and video through one key. These are the latest releases on the gateway — browse and compare all models →
Prepaid credit
Pay only for what you use. There is no subscription and no monthly minimum — buy credit, and each API call draws down your balance at that model's rate.
Pay as you go
0%
service fee — top up once and pay only each model's list rate. No subscription, no minimums.
WEBIE20Top-up packs
You pay the smaller number. You can spend the larger one — bonus is added automatically.
Bonus is added to your usable balance before promo. Add WEBIE20 at checkout for another 20% on your first top-up.
No service fee
Some AI aggregators take a service fee and tax before your money becomes credit. SuperbAPI charges no service fee and adds bonus credit instead.
Same models, one endpoint — and metering you can trust: every call shows its tokens, model, and exact cost in your usage feed, and failed calls are never charged. Third-party fees vary by provider and region.
How billing works
Prepaid credit, spent per call. You only pay for what you use — no monthly bill, no minimum, no surprises.
Top up once
Add credit at any amount. Bonus lands instantly and your balance stays valid for 12 months, across all 84 live models.
Spend per call, per model
Each request deducts that model's rate from your balance. The exact estimated cost is shown live in the workspace before you run.
Failed calls are never charged
If an upstream provider errors, you keep your credit. Pay only for calls that actually return a result.
| Text & reasoning | Billed per token — input + output |
|---|---|
| Image, video & audio | Billed per generated unit |
| Failed upstream calls | Never charged |
| Optional US$20/mo plan | Unlimited keys + 10% off usage |
You only ever spend your prepaid credit. No fixed monthly bill and no minimum — the optional subscription is the only recurring charge, and everything works without it.
Why SuperbAPI
Built so a single prepaid balance and a single base URL replace a drawer full of provider accounts, billing portals, and SDKs.
Text, image, video, and audio from Google and Anthropic behind a single OpenAI-compatible route — switch models per request, never per integration.
Top up once and spend across every model and modality. No per-provider invoices, no monthly minimum — your balance is the only thing that moves.
Issue a separate SuperbAPI key per app or per user, each with its own spend cap, model allowlist, and expiry date.
Your real Google and Anthropic keys live behind SuperbAPI and never reach your users' devices or your client code.
Keep the OpenAI SDK you already use. Change the base URL and key — your existing request and response shapes work unchanged.
When one upstream errors or rate-limits, SuperbAPI retries another automatically — and failed calls are never charged. Top up once and only pay for what you use.
Built for app makers
Issue scoped SuperbAPI keys to your users with per-key spend limits and model rules. Behind them, your real provider keys rotate automatically, so one failing upstream never takes your app down.
Setup guides
Every OpenAI-compatible tool needs the same three values — the base URL, your API key, and an exact model ID. Pick your platform below for a step-by-step walkthrough with the exact fields to fill in and the mistakes that cause errors.
Model catalog
Every model on the gateway in one place. Models tagged Available are callable today with one key and one balance; others are shown for comparison only. The Newest tag flags the latest releases.
Contact & support
Most questions are answered instantly by the SuperbAPI assistant. For account-specific issues, a human follows up within one business day.
Open the chat bubble at the bottom-right of any page and ask about pricing, setup, models, or billing — grounded answers in seconds.
Signed-in users can open a ticket from the workspace Support page. A human admin replies within one business day, and you'll get the reply by email too.
Go to workspace supportPrefer email? Reach the team directly.
support@superbapi.comDeFiner Technology Limited — Hong Kong SAR.
SuperbAPI is an independent AI API gateway and aggregator; third-party model names are used nominatively to indicate routing. See our Terms, Refund Policy, and Privacy Policy.
Workspace access
Prepaid pay-as-you-go: top up once and route it across every major model from one endpoint. No subscription, no service fee — credit is valid for 12 months.
Enter your account email above, then we email a code to confirm it is you before you set a new password.
Use password login, Google, or email code to enter the workspace.
Open verification linkAPI Keys
Create scoped keys, set spend limits, and connect apps through the OpenAI-compatible route.
https://api.aiaccess.local/v1
Bearer YOUR_API_KEY
curl https://superbapi.com/v1/models
await fetch("/v1/chat/completions")
requests.post("/v1/chat/completions")
| Name | Key | Limits | Usage | Status | Last used |
|---|
Estimated US$0
Integrations
One OpenAI-compatible gateway, one balance. Point any SDK here, connect an MCP-native agent, or mint a provisioning token to manage keys from code.
Keep your existing OpenAI SDK. Swap the base URL and pass a SuperbAPI key — chat completions and the model list work unchanged.
https://www.superbapi.com/v1
Bearer YOUR_API_KEY
curl https://www.superbapi.com/v1/models
await fetch("/v1/chat/completions")
requests.post("/v1/chat/completions")
Reference: OpenAPI spec · llms.txt · supported now: /v1/models and /v1/chat/completions, including SSE streaming.
Step-by-step walkthroughs for Cursor, terminal CLIs, and other editors — including the exact fields to fill in and the mistakes that cause “invalid model” errors.
Add SuperbAPI to an MCP-native client — Claude Desktop, Cline, or your own agent — over the Streamable-HTTP transport.
https://www.superbapi.com/mcp
Authorize with Authorization: Bearer YOUR_API_KEY. Tools exposed: list_models and chat.
{
"mcpServers": {
"superbapi": {
"type": "http",
"url": "https://www.superbapi.com/mcp",
"headers": { "Authorization": "Bearer YOUR_API_KEY" }
}
}
}
Mint one token that lets another app or AI create, list, and revoke scoped API keys for your end users — programmatically. It manages keys only; it cannot call models.
No provisioning token yet.
curl -X POST https://www.superbapi.com/api/v1/keys \
-H "Authorization: Bearer YOUR_PROVISIONING_TOKEN" \
-H "Content-Type: application/json" \
-d '{"name":"End-user key","allowedModels":["gpt-4.1-mini"],"budgetLimit":25}'
Endpoints: POST /api/v1/keys (mint), GET /api/v1/keys (list), DELETE /api/v1/keys/:id (revoke).
Models
Browse the full landscape and compare models side by side. Models tagged Available are callable through your key today; others are shown for comparison only. The Newest tag flags the latest releases.
Earn credit
Share your link. Your friend gets +10% extra credit on their first top-up, and you earn 10% of it as credit — released to your balance 14 days later. No limit on how many friends you invite.
Loading…
| Friend | Their top-up | Your credit | Status | Releases |
|---|---|---|---|---|
| No referrals yet. Share your link to start earning. | ||||
Credits & usage
Top up your balance, then track spend, credit lots, and receipts — all in one place. Credit lots are valid for 12 months; promo code WEBIE20 saves an extra 20% on your first top-up.
| Lot | Remaining | Purchased | Expires | Status |
|---|
| Date | Model | Key | Tokens | Cost (est.) | Status |
|---|
Subscription
Pay as you go. Unlock unlimited API key generation and 10% off extra usage.
Support
Hit a problem — errors, slow responses, or a billing question? Report it here first and give us 48 hours to investigate and fix it before opening a payment dispute or chargeback. Our first remedy is usually a fix or a service credit. See the Refund Policy for how refunds work.
Your conversations
Enter to send · Shift+Enter for a new line
Settings
Your signed-in account, local browser state, and legal references.
The account this workspace and its credit belong to.
Clears this browser's cached workspace state (keys and credit shown locally). It does not touch your real account.
How billing, refunds, and your data are handled.
Backend administrators only. Paste your access token to unlock the Admin section (users, provider keys, key groups, usage, and the support inbox).
Admin · Users
Loading users…
| User | Registered | Last active | Balance | Spent | Keys | Status | |
|---|---|---|---|---|---|---|---|
| Loading… | |||||||
Provider Keys
Provider keys are validated against /v1/models. Failed keys are saved as degraded so they do not look healthy.
Key Groups
Groups fill and spill automatically — every user always lands in one with a routable provider key. No manual assignment needed for normal operation.
Admin · Spend monitor
Per-request spend with the account behind each call. Abnormally large calls and outlier accounts are flagged automatically.
| When | Account | Model | Tokens | Cost | Status |
|---|---|---|---|---|---|
| Loading… | |||||
Support Inbox
API key setup
Name the key and apply limits. You can copy the key now and anytime later.
Side-by-side across every attribute. Pick 2–4 models to compare.
https://www.superbapi.com/v1
sk-sbapi-…
claude-fable-5
sk-sbapi-… value (it is shown once).https://www.superbapi.com/v1.sb-claude-fable-5. Cursor reserves built-in IDs like claude-fable-5, so every Claude model also answers to an sb- alias that routes identically. For non-reserved models, type the exact catalog ID — Cursor sends the name you type as the model ID, so a display name like “fable 5” will be rejected upstream.claude-* model and has no base-URL override.Before you file a bug
claude-* IDs), use the sb- alias instead: sb-claude-fable-5. Same model, same price, same routing.GET /v1/models or the Models page.export OPENAI_API_KEY="sk-sbapi-…"
export OPENAI_BASE_URL="https://www.superbapi.com/v1"
curl "$OPENAI_BASE_URL/chat/completions" \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"claude-fable-5","stream":true,
"messages":[{"role":"user","content":"Say OK"}]}'
OPENAI_API_BASE (aider) or OPENAI_BASE_URL (official SDKs) — set both if unsure.Before you file a bug
/v1 — a common failure is pointing at the bare domain.www.superbapi.com as the host. There is no separate api. subdomain.stream: true) is fully supported.https://www.superbapi.com/v1, your sk-sbapi-… key, and the exact model ID.Before you file a bug
/v1; fields labelled “Base URL” usually want it with. If one form fails, try the other.