Pricing

Merius bills per token — you pay for the input and output tokens each request uses, at the model's published rate. No subscriptions, no minimums, no per-request fees.

How billing works

• Per-token. Each model has an input (prompt) price and an output (completion) price, quoted in USD per million tokens. A request costs input_tokens × input_price + output_tokens × output_price.
• Prepaid credits. You add credits in the dashboard; usage draws them down. When your balance reaches zero, requests return 402 until you top up.
• Cached input is cheaper. When a model supports prompt caching, repeated prompt prefixes are billed at the input_cache_read rate — lower than fresh input.

Live prices

Prices are published on the /v1/models endpoint, so they are always current — read them programmatically rather than hard-coding. Each model carries a pricing array with per-token rates as quoted strings:

{
  "id": "z-ai/glm-5.2",
  "pricing": [
    {
      "prompt": "0.0000006",          // $0.60 per 1M input tokens
      "completion": "0.0000022",      // $2.20 per 1M output tokens
      "input_cache_read": "0.0000001" // cached input, per token
    }
  ]
}

Checking your usage

Your current balance and usage are in the dashboard. Every response also includes a usage object with the exact token counts for that request:

"usage": {
  "prompt_tokens": 12,
  "completion_tokens": 97,
  "total_tokens": 109
}

Multiply those counts by the model's per-token prices to compute the exact cost of any request.