{
 "checked": "2026-11-12",
 "units": "USD per 1M tokens, text tokens, short-context Standard tier, list price (not negotiated, not via a reseller)",
 "reuse_note": "Shared file: other KH Labs demos may reuse it. Every number below was read from the provider's own pricing page on the checked date; anything we could not read is listed in 'unverified' instead of being guessed.",
 "sources": {
  "openai": {
   "url": "https://developers.openai.com/api/docs/pricing",
   "checked": "2026-11-12"
  },
  "anthropic": {
   "url": "https://platform.claude.com/docs/en/about-claude/pricing",
   "checked": "2026-11-12"
  },
  "google": {
   "url": "https://aistudio.google.com/docs/pricing",
   "checked": "2026-11-12"
  },
  "deepseek": {
   "url": "https://api-docs.deepseek.com/quick_start/pricing",
   "checked": "2026-11-12"
  },
  "qwen": {
   "url": "https://www.alibabacloud.com/help/en/model-studio/model-pricing",
   "checked": "2026-11-12",
   "note": "International (Singapore) tables"
  },
  "mistral": {
   "url": "https://mistral.ai/pricing",
   "checked": "2026-11-12",
   "note": "Only one API price is printed, in the FAQ"
  }
 },
 "quality_source": {
  "name": "Artificial Analysis Intelligence Index",
  "kind": "third-party composite index (reasoning, knowledge, coding and agentic evals), not the KH Labs benchmark",
  "url": "https://artificialanalysis.ai/leaderboards/models",
  "checked": "2026-11-12",
  "note": "The page does not state an index version or update date. Where several reasoning-effort variants are listed, the 'high' variant is used when it exists; otherwise the nearest listed variant, labelled in aa_variant. Scores are copied as shown (integers)."
 },
 "models": [
  {
   "id": "gpt-6-astra",
   "provider": "OpenAI",
   "name": "GPT-6 Astra",
   "input": 10.0,
   "cached_input": 1.0,
   "output": 50.0,
   "batch_factor": 0.5,
   "aa": 55,
   "aa_variant": "high",
   "note": "Long-context (>272K) rates are higher ($20 in / $75 out)."
  },
  {
   "id": "gpt-6.1-sol",
   "provider": "OpenAI",
   "name": "GPT-6.1 Sol",
   "input": 2.0,
   "cached_input": 0.1,
   "output": 10.0,
   "batch_factor": 0.5,
   "aa": 50,
   "aa_variant": "high",
   "note": ""
  },
  {
   "id": "gpt-6.1-terra",
   "provider": "OpenAI",
   "name": "GPT-6.1 Terra",
   "input": 1.0,
   "cached_input": 0.05,
   "output": 5.0,
   "batch_factor": 0.5,
   "aa": 42,
   "aa_variant": "high",
   "note": ""
  },
  {
   "id": "gpt-6-luna",
   "provider": "OpenAI",
   "name": "GPT-6 Luna",
   "input": 0.1,
   "cached_input": 0.01,
   "output": 0.5,
   "batch_factor": 0.5,
   "aa": 21,
   "aa_variant": "high",
   "note": ""
  },
  {
   "id": "claude-fable-5.2",
   "provider": "Anthropic",
   "name": "Claude Fable 5.2",
   "input": 8.0,
   "cached_input": 0.2,
   "output": 40.0,
   "batch_factor": 0.5,
   "aa": 62,
   "aa_variant": "high",
   "note": "Cache write 5m: $10."
  },
  {
   "id": "claude-opus-5.5",
   "provider": "Anthropic",
   "name": "Claude Opus 5.5",
   "input": 4.0,
   "cached_input": 0.2,
   "output": 20.0,
   "batch_factor": 0.5,
   "aa": 52,
   "aa_variant": "high",
   "note": "Cache write 5m: $5."
  },
  {
   "id": "claude-sonnet-5.5",
   "provider": "Anthropic",
   "name": "Claude Sonnet 5.5",
   "input": 2.0,
   "cached_input": 0.1,
   "output": 10.0,
   "batch_factor": 0.5,
   "aa": 42,
   "aa_variant": "high",
   "note": "Cache write 5m: $2.50."
  },
  {
   "id": "claude-haiku-5.5",
   "provider": "Anthropic",
   "name": "Claude Haiku 5.5",
   "input": 0.1,
   "cached_input": 0.01,
   "output": 0.5,
   "batch_factor": 0.5,
   "aa": 30,
   "aa_variant": "high",
   "note": "Prompts up to 100k tokens; above that $0.50 in / $2.50 out. Cache write 5m: $0.125."
  },
  {
   "id": "gemini-4-carbon",
   "provider": "Google",
   "name": "Gemini 4 Carbon",
   "input": 2.0,
   "cached_input": 0.1,
   "output": 10.0,
   "batch_factor": 0.5,
   "aa": 60,
   "aa_variant": "high",
   "note": "Prices double on 2026-12-01 (input $4.00, cache $0.20, output $20.00)."
  },
  {
   "id": "gemini-4-argon",
   "provider": "Google",
   "name": "Gemini 4 Argon",
   "input": 2.0,
   "cached_input": 0.1,
   "output": 10.0,
   "batch_factor": 0.5,
   "aa": 50,
   "aa_variant": "high",
   "note": "Prices double on 2026-12-01."
  },
  {
   "id": "gemini-3.8-flash",
   "provider": "Google",
   "name": "Gemini 3.8 Flash",
   "input": 0.75,
   "cached_input": 0.075,
   "output": 3.75,
   "batch_factor": 0.5,
   "aa": 31,
   "aa_variant": "high",
   "note": "Prices double on 2027-01-01."
  },
  {
   "id": "gemini-3.5-flash-lite",
   "provider": "Google",
   "name": "Gemini 3.5 Flash-Lite",
   "input": 0.3,
   "cached_input": 0.03,
   "output": 2.5,
   "batch_factor": 0.5,
   "aa": 12,
   "aa_variant": "single listing",
   "note": ""
  },
  {
   "id": "deepseek-v4-pro",
   "provider": "DeepSeek",
   "name": "DeepSeek V4 Pro 0813",
   "input": 1.32,
   "cached_input": 0.044,
   "output": 3.96,
   "batch_factor": null,
   "aa": 23,
   "aa_variant": "max",
   "note": "Peak-hour price shown (01:00-04:00 and 06:00-10:00 UTC, Mon-Fri); off-peak is exactly half ($0.66 / $0.022 / $1.98). No batch discount printed."
  },
  {
   "id": "deepseek-flash",
   "provider": "DeepSeek",
   "name": "DeepSeek V4.1 Flash 1017",
   "input": 0.3,
   "cached_input": 0.006,
   "output": 1.2,
   "batch_factor": null,
   "aa": 29,
   "aa_variant": "high",
   "note": "Peak-hour price shown; off-peak is half ($0.15 / $0.003 / $0.60). No batch discount printed."
  },
  {
   "id": "qwen3.8-max",
   "provider": "Qwen (Alibaba Cloud)",
   "name": "Qwen3.8 Max",
   "input": 2.0,
   "cached_input": null,
   "output": 6.0,
   "batch_factor": null,
   "aa": 34,
   "aa_variant": "0902 (single listing)",
   "note": "International (Singapore) price, 0-1M tier. A context-cache discount exists but its price for this model is not printed on the page, so it is not modelled. Artificial Analysis lists 'Qwen3.8 Max (0902)'; the pricing page lists 'qwen3.8-max' without a date, so the match is by name."
  },
  {
   "id": "qwen3.8-27b",
   "provider": "Qwen (Alibaba Cloud)",
   "name": "Qwen3.8 27B (hosted)",
   "input": 0.5,
   "cached_input": null,
   "output": 3.0,
   "batch_factor": null,
   "aa": 20,
   "aa_variant": "xhigh",
   "note": "International price. Cache discount not printed."
  }
 ],
 "unverified": [
  "LMArena / arena.ai text leaderboard: the page returned HTTP 403 to our fetch, so no Arena scores are used anywhere.",
  "Llama via hosted providers: Together's pricing page lists Llama only for fine-tuning and its serverless catalog lists no Llama model; Groq's pricing page prints no per-token prices. No verified Llama price, so Llama is not included. Meta's current models on Artificial Analysis (Muse family) have no API price we could verify.",
  "Mistral Medium, Small, Ministral and Codestral API prices: not printed on mistral.ai/pricing or the models overview. They have Artificial Analysis scores but cannot be costed.",
  "Qwen context-cache prices: the Alibaba page prints one only for qwen3.8-omni-flash. Its generic Context Cache example (hits at 10% of input) was not applied to other models.",
  "Qwen3.8 Max version match between the pricing page (no date) and Artificial Analysis (0902) is by name only.",
  "Artificial Analysis does not state an index version or update date on the page; 2026-11-12 is when we read it.",
  "No Artificial Analysis score was found for GPT-5.4 mini/nano, Gemini 3.1 Pro, Gemini 3.1 Flash-Lite, Gemini 3 Flash, Qwen3.8 Flash. Those rows show price only.",
  "Batch prices for DeepSeek, Qwen3.8 and Gemini 3.1 Pro were not printed on the pages we read, so no batch variant is shown for them.",
  "Removed from the page because a verified Artificial Analysis score or official price is missing: GPT-5.4 mini, GPT-5.4 nano, Gemini 3.1 Pro (preview), Gemini 3.1 Flash-Lite, Gemini 3 Flash (preview), Qwen3.8 Flash, Mistral Large (version not stated). Gemini 3.8 Flash (tiered), Gemini 3.1 Flash image and gpt-oss are not on the page either."
 ],
 "removed_for_missing_score_or_price": [
  "GPT-5.4 mini",
  "GPT-5.4 nano",
  "Gemini 3.1 Pro (preview)",
  "Gemini 3.1 Flash-Lite",
  "Gemini 3 Flash (preview)",
  "Qwen3.8 Flash",
  "Mistral Large (version not stated)"
 ]
}