Your community is your API key.
Top up once. Send tokens to anyone. They spend them across Claude, GPT-5, Gemini, Mistral and every other frontier model — one balance, every lab.
Top up once. Send tokens to anyone. They spend them across Claude, GPT-5, Gemini, Mistral and every other frontier model — one balance, every lab.
Send Me A Token sits between you and every frontier model. You don't need their key. You need your community.
Load a balance with a card, ACH, or a friend who got there first. No per-seat licensing, no team plan, no “contact sales.” The balance sits in your wallet until you spend it.
Pick a handle. Choose an amount. Hit send. Transfers settle peer-to-peer in about a second, with an optional note attached and a receipt at both ends.
Recipients use their balance across Claude, GPT-5, Gemini, Mistral, Llama and more. Pick the right model per message, not per contract. One balance, one auth, every lab.
Frontier model access is gated behind seat licenses, team plans, and brittle keys. Most people don't need an account with five labs. They need one balance, one log-in, and a friend who'll spot them.
You never sign up for Anthropic, OpenAI, Google or anyone else. Send Me A Token carries the relationship; you carry your community.
Every send, top-up and spend has a public-ish receipt: copyable tx id, note, settle time. Auditable without being noisy.
A new wallet is one share-link away. The person you sent 100 tokens to is signed up by the time the receipt hits your inbox.
Hackathon team, class cohort, open-source maintainers. Point a shared balance at the same handle and split the work.
Your balance becomes an API key. Point Cursor, Claude Code, Hermes Agent, Open Code — any agentic or CLI tool — at it. We route to whichever frontier model your tokens can pay for. You never juggle vendor keys again.
bmt_live_••••_a2f7Spend tokens from the web app, from your terminal, or from your code. One key, every model, no per-vendor SDK.
# spend tokens against ANY model. one endpoint. curl https://api.buymetoken.dev/v1/chat \ -H "Authorization: Bearer bmt_live_…" \ -d '{ "model": "claude-opus-4", "input": "summarise the manifesto in 3 lines.", "max_tokens": 256 }' # response "tokens_spent": 142, "balance": 9858, "settled_in_ms": 1208, "vendor_key_used": false
// spend tokens against ANY model. one endpoint. import { BMT } from "@buymetoken/sdk"; const bmt = new BMT({ apiKey: process.env.BMT_KEY }); const { output, tokens_spent, balance } = await bmt.chat({ model: "claude-opus-4", input: "summarise the manifesto in 3 lines.", max_tokens: 256, }); console.log({ tokens_spent, balance }); // → { tokens_spent: 142, balance: 9858 }
# spend tokens against ANY model. one endpoint. from buymetoken import BMT bmt = BMT(api_key=os.environ["BMT_KEY"]) resp = bmt.chat( model="claude-opus-4", input="summarise the manifesto in 3 lines.", max_tokens=256, ) print(resp.tokens_spent, resp.balance) # → 142 9858
Account, sending, every model — all $0. We add a small transaction fee on top-ups to cover card processing and routing. That's the entire pricing page.
Pure math, live rates, no hidden margin. Every USD you upload is converted at the cheapest applicable model price.
Rate updates live as labs adjust their per-token prices. You see the math before you confirm. No subscriptions, no seats, no minimums.
Ours. Send Me A Token carries the vendor relationship with every model lab on your behalf, including Anthropic, OpenAI, Google, Mistral and the rest. You spend tokens against your balance; we settle with the lab in the background. You never see a vendor key, never get rate-limited by one, never lose access because somebody rotated something.
You can. Crew accounts support BYO keys for compliance reasons, but the whole point of Send Me A Token is that you don't have to. The product is built around not needing one.
No. Tokens are prepaid usage credits, denominated in inference, not currency. They're non-transferable to fiat, non-redeemable for cash, and they don't earn interest. They are transferable between Send Me A Token wallets, which is the entire premise.
Your balance is independent of any single vendor. If Anthropic has an outage, you route to GPT-5 or Gemini for that prompt, with the same key, same wallet, no migration. That's the upside of one balance across every model.
Claude (Opus, Sonnet, Haiku), GPT-5, Gemini 2.5 Pro, Mistral Large, Llama 3.3, DeepSeek V3, Grok 3, Qwen 3, Cohere Command and Perplexity Sonar. We add new frontier releases within days of public availability — your existing balance picks them up automatically, no contract renegotiation.
Yes. Any tool that accepts an OpenAI-compatible API key works out of the box — Cursor, Claude Code, Codex, Cline, Antigravity, Junie, Hermes Agent, Open Code, and the rest. Drop your bmt_live_… key in their settings, pick a model, spend from your balance.
When you top up in USD, we deduct a 5% transaction fee, then convert the rest to tokens at the live blended model rate. There's no markup on inference itself — you spend tokens against published per-million rates, and the math is shown to you before every top-up. No subscriptions, no seats, no minimums.
Yes. All traffic is TLS-terminated at our edge, prompts are not logged after dispatch, and per-key rate limits + spend caps keep a runaway loop from draining your balance. Crew accounts add audit logging, SSO, and per-member spend limits.
We don't train on them, we don't sell them, and we don't hand them to third parties. Prompts pass through to the model lab you selected and the response comes back — we keep only the bytes-in / bytes-out for billing, plus a short-window error log for debugging. That's it.
Tokens never expire, so most people don't need to. If you do — within 30 days of a top-up, message support and we'll refund the unspent USD-equivalent minus the original transaction fee. Beyond 30 days, balances stay live indefinitely on your account.