Skip to content
XPNCORE

One API key for every AI model

Every top AI model, one API

Call GPT, Claude, Gemini and more with one API key that speaks the OpenAI, Anthropic and Gemini APIs. Pay only for what you use.

Sign up freeBrowse modelsRead the docs

Already have an account?Open dashboard

Example request log

Example of one API key's latest requests
ModelLatencyCostCode
claude-sonnet-4-53.8 s₫2,017200
gpt-5-mini1.1 s₫47200
gemini-2.5-pro6.2 s₫896200
claude-sonnet-4-50.4 s₫0429

Illustrative figures. Failed requests are never charged.

# Paste into a terminal (macOS, Linux)export ANTHROPIC_BASE_URL="https://api.xpncore.com"export ANTHROPIC_AUTH_TOKEN="YOUR_API_KEY"# Then run it in your project folderclaude
After signing up, the Connect page writes a one-line setup with your key and models.Model ids:GET /v1/models

  • 51models

    One key for every model

    Models with a listed price in the Models price table

  • 4API formats

    OpenAI, Anthropic, Gemini, Ollama

    Change the base URL and go

  • ₫0

    Failed requests are never charged

    Retries and failovers are charged once

  • 30days

    A log of every request

    Tokens, cost, latency, error codes

Features

Why XPNCORE

A thin API layer between your app and AI models: steady, transparent and easy on costs.

Read the API docs

  • One key, every model

    Use a single API key to call AI models from OpenAI, Anthropic, Google and many more.

    base_url = https://api.xpncore.com/v1

  • Automatic failover

    If one source has a problem, requests move to another so your app keeps running.

    503 → backup source → 200

  • Live monitoring

    See requests, tokens, cost and error rate in real time.

    requests · tokens · cost · latency

  • Security

    Encrypted API keys, IP limits, two-factor authentication and passkeys.

    2FA · passkeys · IP limits

  • Model combos

    Group several models under one id of your own. Each request is served by one model in the combo and charged at that model's price.

    cb-<name> = model A → model B

  • Per-key limits

    Cap each key's spend per day, week or month, and its tokens. A key that hits its cap stops; your other keys keep running.

    day · week · month · total · tokens

Connect

Connect your tools with one command

The Connect page writes the setup command or config for each tool, with your key and models filled in.

Open the Connect page
  • Claude Code CLI1 command
  • ChatGPT Codex CLI / App1 command
  • OpenCode1 command
  • Factory Droid1 command
  • Crush1 command
  • Claude Desktopready config
  • Cline / Roo Codeready config
  • Cherry Studioready config
  • LobeChatready config
  • Continueready config
  • CC Switchready config
  • Cursorready config
  • Any other appSDKs, OpenAI formatRead the API docs

On the web

More than an API

Use it in the browser with the same account, no code needed.

  • Chat AI

    Chat with any model and send images, files and voice from the browser.

    Images · files · voice

  • Image generation

    Create images from a prompt or edit your own.

    Paid per image

  • Video generation

    Short videos from text or an image.

    Paid per video

  • Playground

    Try models and settings before you write code.

    Side-by-side compare

Pricing

A public price for every model

Prices per 1M tokens; you pay for what you use. Failed requests are never charged.

Some models and their current prices
ModelInput / 1M tokensOutput / 1M tokens
gpt-6.1-solIn₫3,000Out₫10,000
claude-sonnet-5-5In₫2,400Out₫10,000
claude-opus-5-5In₫4,000Out₫15,000
claude-opus-5.5-cheapIn₫1,800Out₫4,000
claude-opus-4-6In₫1,000Out₫5,000
gpt-6-lunaIn₫500Out₫1,000
The 6 models with the most tokens in 30 days · 51 priced publiclySee every price

Billing

Pay only for what you use

Each request is deducted from your wallet at the price of the model it used.

How charges work

Successful request
By input and output tokens
Failed request
Not charged
Retries and failovers
Charged once
Out of balance or over a limit
Refused, not charged
  • Set a spending cap for each API key.
  • See each request's cost in the request log.
  • Balances show in the currency you choose.

FAQ

Frequently asked questions

Am I charged more than once when a request fails or is retried?

No. However many times we retry internally or switch sources, you pay once for the final result. Failed requests are not charged.

Does XPNCORE work with the OpenAI, Anthropic and Gemini SDKs?

Yes. XPNCORE accepts requests in the OpenAI (Chat Completions, Responses), Anthropic Messages (Claude Code, Anthropic SDKs), Google Gemini and Ollama formats. Change the base URL and API key; everything else stays the same.

Do I need a separate API key for each model maker?

No. One XPNCORE API key works for every model you can access, from OpenAI, Anthropic, Google and the other models in the catalog.

How do I know which models my account can use?

Call GET /v1/models with your API key, or open the Models & Pricing page. Use the exact model ID from the list when you send a request.

See all questions

Community

Join the user community

Ask how to connect your tools, and hear first about new models, maintenance and offers.

Get started

Ready to get started?

Sign up for free and start using APIs today.

Get StartedRead the docs
  1. Sign upFree
  2. Create an API keyDashboard › API keys
  3. Change the base URLhttps://api.xpncore.com/v1