base-context/benchmarks/python-realworld-30/native/pricing-metadata.json at main · BaseModelAI/base-context · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
{
  "schema": "published-api-pricing/v1",
  "recorded_on": "2026-10-03",
  "currency": "USD",
  "unit": "million_tokens",
  "basis": "Official published API list rates, not invoices or subscription cash charges.",
  "models": {
    "gpt-6.1-sol": {
      "provider_family": "openai",
      "input": 2.0,
      "cache_read": 0.1,
      "cache_write": 2.5,
      "output": 10.0,
      "source": "https://developers.openai.com/api/docs/models/gpt-6.1-sol"
    },
    "gpt-6-astra": {
      "provider_family": "openai",
      "input": 10.0,
      "cache_read": 1.0,
      "cache_write": 12.5,
      "output": 50.0,
      "source": "https://developers.openai.com/api/docs/models/gpt-6-astra"
    }
  },
  "model_aliases": {},
  "openai_modifiers": {
    "long_input_above_tokens": 272000,
    "long_input_multipliers": {
      "input": 2,
      "cache_read": 2,
      "cache_write": 2,
      "output": 1.5
    },
    "tier_multipliers": {
      "default": 1,
      "standard": 1,
      "fast": 2,
      "priority": 2,
      "flex": 0.5,
      "batch": 0.5
    },
    "unknown_tier": "Assume standard list rate and mark this explicitly; do not claim actual invoiced tier."
  },
  "limitations": [
    "Unsupported models remain unpriced. No manifest-selected model is substituted without request metadata.",
    "Reasoning tokens are a subset of output; never add reasoning output a second time.",
    "Per-record costs concern captured native requests/responses only. Family coverage can remain unknown.",
    "No regional-processing modifier is inferred; only supplied explicit service-tier and prompt-length rules are applied."
  ],
  "collector_input_semantics": {
    "base-context": "attempts[].usage.input is uncached ordinary input; inputTotal includes cache categories. Only native attempt records are priced.",
    "codex": "response_receipts[].usage.input_tokens includes cache read/write categories; subtract reported cache categories before pricing ordinary input. Model comes from source turn_context, not verified backend resolution.",
    "reasoning": "reasoning_output_tokens is already inside output_tokens; not added."
  },
  "timestamp_caveat": "Base uses sentAt/admittedAt when recorded. Codex rollout timestamp is only a timestamp proxy, not exact provider dispatch time."
}