Meter Documentation
Introduction
Meter is the first consumer AI product with postpaid billing. No subscription. No credits. Use first, pay after. The meter runs up in dollars like a taxi.
Every model available — Claude, GPT, Gemini, Grok, DeepSeek. One bill. No complexity.
How It Works
- Sign up with your email
- Add a card — no charge, just a verification hold
- Start chatting — every model is available
- Each response shows:
Model · $Cost · Confidence% - Your card is charged at $10 or monthly, whichever comes first
Pay Per Use
Every message has a cost based on the model used and tokens consumed. Below each response you see:
Model name. Cost in dollars. AI confidence score. Settlement status.
Pricing
Pay-per-token at provider base rates (no markup by default). Most messages cost between $0.01 and $0.25.
| Model | Input / 1M | Output / 1M | ~Per Msg |
|---|---|---|---|
| Claude Sonnet 4.6 | $6.00 | $30.00 | ~$0.02 |
| Claude Opus 4.6 | $10.00 | $50.00 | ~$0.03 |
| GPT-5.4 | $5.00 | $30.00 | ~$0.02 |
| MiniMax M2.5 | $0.60 | $2.40 | ~$0.001 |
| Gemini 3.1 Pro | $4.00 | $24.00 | ~$0.01 |
| Grok 4.1 Fast | $0.40 | $1.00 | ~$0.001 |
| DeepSeek V3 | $0.54 | $2.20 | ~$0.001 |
| Meter 1.0 (Debate) | $13.90 | $79.00 | ~$1.00 |
The daily meter in the header shows your running total. Set a daily spending cap in settings.
Billing
Your card is charged when your balance reaches $10, or at the end of each month — whichever comes first. Stripe handles all payments securely.
Most founders spend $20-40/month — less than a single AI subscription, with every model included.
Models
Every frontier model available through one interface. Quality measured by GPQA Diamond (graduate-level science accuracy). Speed is output tokens per second from native APIs.
| Model | Provider | GPQA | Speed |
|---|---|---|---|
| Claude Sonnet 4.6 | Anthropic | 74% | 60 tok/s |
| Claude Opus 4.6 | Anthropic | 91% | 70 tok/s |
| GPT-5.4 | OpenAI | 92% | 90 tok/s |
| MiniMax M2.5 | MiniMax | 62% | 50 tok/s |
| Gemini 3.1 Pro | 92% | 138 tok/s | |
| Grok 4.1 Fast | xAI | 86% | 129 tok/s |
| DeepSeek V3 | DeepSeek | 59% | 50 tok/s |
| Meter 1.0 (Debate) | Meter | 93% | 30 tok/s |
Meter 1.0 runs Opus + GPT-5.4 + Grok in a multi-model debate, then synthesizes the best answer. Slower but highest quality.
API Reference
POST /api/v1/chat
Streaming chat endpoint. Returns SSE events.
POST /api/v1/chat
Authorization: Bearer mk_your_api_key
Content-Type: application/json
{
"messages": [{"role": "user", "content": "Hello"}],
"model": "anthropic/claude-sonnet-4.6"
}
// Response: SSE stream
data: {"type":"delta","content":"Hi","tokensOut":1}
data: {"type":"usage","tokensIn":5,"tokensOut":50,"confidence":85}
data: {"type":"done"}