{
  "schema": 1,
  "as_of": "2026-11-01",
  "currency": "USD",
  "unit": "USD per million tokens",
  "policy": "Standard real-time API list prices; fixed DeepSeek off-peak rates for all requests, Xiaomi overseas USD rates. No time-of-day switching, batch, account, or regional adjustments.",
  "scope": "Selected coding agent model requests only; excludes verifier, task-internal APIs, infrastructure and unrelated runs. Recorded cache reads and writes are subsets of input. Reasoning is already included in output.",
  "models": {
    "openai/gpt-6-luna": {
      "official_model": "gpt-6-luna",
      "source": "https://developers.openai.com/api/docs/models/gpt-6-luna",
      "rates": {"input": "0.10", "cached_input": "0.01", "cache_write_input": "0.125", "output": "0.50"},
      "long_context": {
        "above_input_tokens": 272000,
        "rates": {"input": "0.20", "cached_input": "0.02", "cache_write_input": "0.25", "output": "0.75"}
      },
      "note": "Standard processing. Prompts above 272,000 input tokens use long-context rates for the entire request. Cache writes use the standard automatic-cache write rate."
    },
    "deepseek/deepseek-v4.1-flash-1017": {
      "official_model": "deepseek-flash (DeepSeek-V4.1-Flash-1017)",
      "source": "https://api-docs.deepseek.com/quick_start/pricing/",
      "rates": {"input": "0.15", "cached_input": "0.003", "cache_write_input": "0.15", "output": "0.60"},
      "note": "Fixed official off-peak rates for every request, independent of run time. Cache misses, including writes, are charged at the input rate without a separate write surcharge."
    },
    "xiaomi/mimo-v2.6-flash": {
      "official_model": "mimo-v2.6-flash",
      "source": "https://mimo.mi.com/docs/en-US/price/pay-as-you-go",
      "rates": {"input": "0.14", "cached_input": "0.0028", "cache_write_input": "0.14", "output": "0.28"},
      "note": "Overseas real-time API prices in USD. Cache writes have no additional charge; their input processing is charged at the cache-miss rate."
    }
  }
}