{
 "as_of": "2026-10-13",
 "verdicts": [
  {
   "job": "Best overall reasoning",
   "method": "Highest Artificial Analysis Intelligence Index (independent).",
   "winner": "Claude Opus 5.5",
   "company": "Anthropic",
   "evidence": "Scores 58 at max effort. Next: Claude Sonnet 5.5 56, Muse Wonder Full 56, Claude Fable 5.1 53.",
   "sources": [
    "aa_lb"
   ],
   "as_of": "2026-10-13",
   "runners": [
    "Claude Sonnet 5.5",
    "Muse Wonder Full",
    "Claude Fable 5.1"
   ],
   "caveat": "At that setting it costs $5.98 per index task and waits 691.7s before the first token."
  },
  {
   "job": "Best coding",
   "method": "Highest Terminal-Bench 4.0 resolution rate — the coding test inside the AA index.",
   "winner": "Claude Sonnet 5.5",
   "company": "Anthropic",
   "evidence": "70.6%. Next: Claude Opus 5.5 66.4%, GPT-6 Astra 57.9%, Claude Fable 5.1 55.8%.",
   "sources": [
    "anth_fable",
    "anth_opus",
    "anth_sonnet"
   ],
   "as_of": "2026-10-13",
   "runners": [
    "Claude Opus 5.5",
    "GPT-6 Astra",
    "Claude Fable 5.1"
   ],
   "caveat": "Every Terminal-Bench figure here is vendor-reported; we could not load an independent coding leaderboard on this date. On a different test, DeepSWE v1.1, Gemini 4 Argon leads at 77.9% (also vendor-reported) — so this verdict is contested."
  },
  {
   "job": "Best value per dollar",
   "method": "Most AA index points per dollar of AA cost-per-task, among settings scoring 40 or higher.",
   "winner": "MiMo-V2.6-Pro",
   "company": "Xiaomi (MiMo)",
   "evidence": "Index 46 at $0.13 per task = 354 points per dollar. Next: Claude Haiku 5.5 (max) 205, GPT-6.1 Sol (high) 156, Claude Opus 5.5 (medium) 38.",
   "sources": [
    "aa_mimo"
   ],
   "as_of": "2026-10-13",
   "runners": [
    "Claude Haiku 5.5 (max) 205",
    "GPT-6.1 Sol (high) 156",
    "Claude Opus 5.5 (medium) 38"
   ],
   "caveat": "Among the eleven major labs the pick is Claude Haiku 5.5 (max): index 43 at $0.21 per task (205 points per dollar)."
  },
  {
   "job": "Fastest",
   "method": "Highest measured output tokens per second among settings scoring 40 or higher on the AA index.",
   "winner": "Muse Spark 1.3 (xhigh)",
   "company": "Meta",
   "evidence": "245 tokens/sec at index 45, first token after 22.3s. Next: Claude Haiku 5.5 (max) 237.9, Gemini 3.8 Flash (high) 141.4, Claude Sonnet 5.5 (max) 135.",
   "sources": [
    "aa_lb"
   ],
   "as_of": "2026-10-13",
   "runners": [
    "Claude Haiku 5.5 (max) 237.9",
    "Gemini 3.8 Flash (high) 141.4",
    "Claude Sonnet 5.5 (max) 135"
   ],
   "caveat": "Raw speed champion regardless of intelligence: Celeris-1 at 1,522 tokens/sec, index 6."
  },
  {
   "job": "Best open-weight",
   "method": "Highest AA Intelligence Index among models whose weights are published.",
   "winner": "MiMo-V2.6-Pro",
   "company": "Xiaomi (MiMo)",
   "evidence": "Index 46, licence: MIT. Next: GLM-5.3 45 (GLM-5.3 License (commercial use with restrictions)), Kimi K3 44 (Kimi K3 License (commercial use with restrictions)), DeepSeek V4.1 Flash 39 (MIT).",
   "sources": [
    "aa_mimo"
   ],
   "as_of": "2026-10-13",
   "runners": [
    "GLM-5.3",
    "Kimi K3",
    "DeepSeek V4.1 Flash"
   ],
   "caveat": "Open weights means you may host it yourself; check each licence — several allow commercial use only with restrictions."
  },
  {
   "job": "Best for voice / agents",
   "method": "Highest Artificial Analysis Speech-to-Speech Index (reasoning, conversation flow and tool use, measured by voice).",
   "winner": "GPT-Live-1 (Astra, medium)",
   "company": "OpenAI",
   "evidence": "Index 83.2, 74.5% on the agentic voice test, first audio in 1.34s, about $5.83 per hour. Next: Gemini 3.8 Live Extended Thinking (High) 82.6, Grok Voice Think Fast 2.0 High 81.3. Cheapest in the top six: Gemini 3.8 Live at $0.84 per hour.",
   "sources": [
    "aa_sts"
   ],
   "as_of": "2026-10-13",
   "runners": [
    "Gemini 3.8 Live Extended Thinking (High)",
    "Grok Voice Think Fast 2.0 High"
   ],
   "caveat": "For text agents doing office work, Claude Opus 5.5 leads GDPval-AA at 1,846 Elo, ahead of Claude Sonnet 5.5 at 1,844."
  }
 ],
 "picker": [
  {
   "id": "think",
   "job": "Hard thinking — contracts, strategy, numbers",
   "pick": "Claude Opus 5.5",
   "why": "Top of the AA Intelligence Index at 58; $4 in / $20 out per million tokens.",
   "plain": "Use the smartest model you can get. Mistakes here cost more than tokens."
  },
  {
   "id": "write",
   "job": "Write emails, posts and proposals in my voice",
   "pick": "Claude Sonnet 5.5",
   "why": "Highest index (56) among models at $10 or less per million output tokens.",
   "plain": "Near-top intelligence at a mid-tier price is the sweet spot for daily writing."
  },
  {
   "id": "code",
   "job": "Build or fix software and automations",
   "pick": "Claude Sonnet 5.5",
   "why": "70.6% on Terminal-Bench 4.0 (vendor-reported); $2 in / $10 out.",
   "plain": "Pick on a coding benchmark, not on general IQ."
  },
  {
   "id": "volume",
   "job": "Answer a high volume of chats, texts or reviews",
   "pick": "Claude Haiku 5.5",
   "why": "Index 43 at $0.21 per index task; $0.10 in / $0.50 out.",
   "plain": "You want the most intelligence per dollar, from one of the major labs."
  },
  {
   "id": "voice",
   "job": "Answer the phone with a voice agent",
   "pick": "GPT-Live-1 (Astra, medium)",
   "why": "Top of the Speech-to-Speech Index at 83.2; about $5.83 per hour of audio.",
   "plain": "Voice needs its own model class — judge on a speech-to-speech benchmark."
  },
  {
   "id": "instant",
   "job": "Instant answers inside a live screen or widget",
   "pick": "Muse Spark 1.3",
   "why": "245 tokens/sec at xhigh effort while still scoring 45.",
   "plain": "Speed first, as long as the model is still capable."
  },
  {
   "id": "private",
   "job": "Keep customer data on my own hardware",
   "pick": "MiMo-V2.6-Pro",
   "why": "Index 46 under the MIT licence.",
   "plain": "Open weights under a permissive licence, so you can host it yourself."
  }
 ]
}