Introduction to MintRouter

MintRouter is an API router and gateway for large language models. One API key gives you access to many models across the OpenAI-compatible, Anthropic-compatible, and Gemini-compatible wire formats. The SDKs and tools you already use keep working — only the base URL and the key change, no code changes required.

Behind a single endpoint, MintRouter routes each request to the right upstream model, meters every token, and gives you one unified place to track usage and cost.

Why use MintRouter?#

One key, one endpoint, many models#

Instead of managing separate accounts, keys, and billing for each provider, you point your client at MintRouter once. The same key can call claude-sonnet-4-6, gpt5.4, gemini-3.1-pro-preview, kimi-k2.7, and every other model in the catalog — switching models is a one-line change to the model field.

Predictable, transparent cost#

Pricing is quoted in USD per 1M tokens, so the cost of a request is easy to reason about before you send it. Per-model rates are published on the /pricing and /models pages, and every metered request is itemized in your dashboard.

Quotas and usage visibility#

Spending caps stop runaway usage before it becomes a surprise. Every request is logged with its model, token counts, and cost, so you always know what was spent, on which model, and when. See /logs and /analytics after signing in.

Compatibility matrix#

MintRouter is a drop-in replacement: keep your existing SDK and swap the base URL.

Key endpoints per format:

• OpenAI format — POST /v1/chat/completions, POST /v1/completions, POST /v1/responses, GET /v1/models. Auth via Authorization: Bearer <key>.
• Anthropic format — POST /v1/messages, POST /v1/messages/count_tokens. Auth via the x-api-key: <key> header.
• Gemini format — POST /v1beta/models/{model}:generateContent and :streamGenerateContent.

Supported features#

The full request/response surface of each format is supported natively, including:

• Streaming (server-sent events)
• Vision / image input
• Function calling / tool use
• Extended thinking / reasoning mode
• Prompt caching (Anthropic format)

Model catalog#

MintRouter serves current frontier models — for example claude-opus-4-8, claude-sonnet-4-6, claude-haiku-4-5-20251001, gpt5.4, gpt5.4-mini, gemini-3.1-pro-preview, kimi-k2.7, and fable-5. Short aliases such as opus4.8, sonnet4.6, and haiku4.5 are also accepted.

The authoritative list is always available two ways:

curl https://api.mintrouter.ai/v1/models \
  -H "Authorization: Bearer $MINTROUTER_API_KEY"

or on the /models page, which also shows per-model pricing.

How it works#

1. Register at https://mintrouter.ai/register.
2. Add balance at /topup, or pick a plan at /plans.
3. Create an API key on the /api-keys page of your dashboard.
4. Point your SDK or tool at the MintRouter base URL for the format you use, with your key as the credential.
5. Send requests as usual — MintRouter routes them, meters the tokens, and deducts the cost from your balance.
6. Track usage at /logs and /analytics.

Top up your balance#

MintRouter is prepaid: requests draw from the USD balance on your account. There are two ways to fund it:

• USDT deposit (pay-as-you-go) — the /topup page shows a merchant address plus a personal memo. Send USDT on the TON network from any exchange (Binance, OKX, Bybit, …) or wallet app — no crypto wallet setup needed. The deposit is detected on-chain and credited to your balance automatically, usually within minutes.
• Plans — prefer a predictable subscription? Pick a plan at /plans instead of pay-as-you-go.

Deposit bonus promotions run from time to time — when one is active it is shown directly on the /topup page. Your current balance and every charge are listed at /billing.

Where to next#

• Quickstart — from account creation to your first request in a few minutes.
• OpenAI Format — full integration guide for OpenAI-compatible SDKs and tools.
• Anthropic Format — the /v1/messages API, including tool use, extended thinking, and prompt caching.
• /models — the model catalog with per-model pricing.
• /pricing — plans and USD-per-1M-token rates.

If you hit an error, check the Quickstart troubleshooting table first — most issues are an invalid key, an unknown model id, or an empty balance.

Contact & support#

Support is 24/7 — a real person answers on both channels, around the clock:

• Email: [email protected] — best for account, billing, and technical questions. Include the email you signed up with so we can find your account quickly.
• Telegram: @mintrouteropen_in_new — quick questions and community chat.

When reporting a problem, include what you were trying to do, what happened instead, and any error message or request ID from your dashboard — it makes the first reply much faster. For billing disputes, account suspension, or data requests, email is the fastest channel; the terms that govern your account are on the Policy page.