{
 "claim": "Jev + Mercury 2.5 over WebMCP is 16.3x to 245.1x cheaper per attempt than the eleven screen- and page-driving configurations on WindTunnel's canonical board; the median of that range is 58.9x. The published 245x is the top of the range - GPT-6 Astra, the priciest model on the board, on screenshot computer use. The cheapest screen-driving configuration, GPT-5.6 Luna on computer use, is 16.3x.",
 "method": "Ratio of medians of est_cost_usd over each configuration's 147 attempts, taken from results/canonical/results.csv at commit 5ca8644. 'Blended $/Mtok' is that configuration's total est_cost_usd divided by its total processed tokens (uncached input + cache reads + cache writes + output). Solved is majority-of-three, as the board reports it. I recomputed est_cost_usd from the four token columns and harness/lib.mjs PRICES for all three GPT-6 Astra rows; all three medians reproduce to the cent.",
 "source": "https://github.com/nekuda-ai/WindTunnel",
 "captured": "2026-09-19",
 "note": "est_cost_usd counts model tokens at pinned list prices and nothing else - no browser, no Docker capsule, no site hosting, and none of the engineering that puts the tools on the page. Jev is priced at $0.042/M for every token type with output free; Mercury at $0.20/$0.02/$0.75 per M (input/cached/output). A 5x Mercury promotional discount was recorded and deliberately not applied.",
 "columns": [
  {
   "key": "cfg",
   "label": "configuration"
  },
  {
   "key": "iface",
   "label": "interface"
  },
  {
   "key": "solved",
   "label": "solved",
   "align": "right"
  },
  {
   "key": "medcost",
   "label": "median $/attempt",
   "align": "right"
  },
  {
   "key": "medtok",
   "label": "median tokens",
   "align": "right"
  },
  {
   "key": "blended",
   "label": "blended $/Mtok",
   "align": "right"
  },
  {
   "key": "ratio",
   "label": "x Jev+Mercury",
   "align": "right"
  }
 ],
 "rows": [
  {
   "cfg": "Jev + Mercury 2.5",
   "iface": "DOM (ultrafast)",
   "solved": "25/49",
   "medcost": "$0.000778",
   "medtok": "13,892",
   "blended": "$0.050",
   "ratio": "0.7x"
  },
  {
   "cfg": "Jev + Mercury 2.5",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.001064",
   "medtok": "9,793",
   "blended": "$0.109",
   "ratio": "\u2014"
  },
  {
   "cfg": "GPT-5.6 Luna",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.002436",
   "medtok": "2,596",
   "blended": "$0.846",
   "ratio": "2.3x"
  },
  {
   "cfg": "Gemini 3.6 Flash \u00b7 Stagehand v4",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.004372",
   "medtok": "4,371",
   "blended": "$0.997",
   "ratio": "4.1x"
  },
  {
   "cfg": "Gemini 3.6 Flash",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.004396",
   "medtok": "4,453",
   "blended": "$1.002",
   "ratio": "4.1x"
  },
  {
   "cfg": "Sonnet 5",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.009132",
   "medtok": "5,172",
   "blended": "$1.716",
   "ratio": "8.6x"
  },
  {
   "cfg": "Sonnet 5 \u00b7 Stagehand v4",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.009511",
   "medtok": "5,161",
   "blended": "$1.732",
   "ratio": "8.9x"
  },
  {
   "cfg": "GPT-5.6 SOL",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.012100",
   "medtok": "2,573",
   "blended": "$4.239",
   "ratio": "11.4x"
  },
  {
   "cfg": "Claude Opus 5",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.014013",
   "medtok": "4,770",
   "blended": "$3.472",
   "ratio": "13.2x"
  },
  {
   "cfg": "GPT-6 Astra",
   "iface": "WebMCP",
   "solved": "49/49",
   "medcost": "$0.017100",
   "medtok": "2,575",
   "blended": "$7.006",
   "ratio": "16.1x"
  },
  {
   "cfg": "GPT-5.6 Luna",
   "iface": "computer use",
   "solved": "45/49",
   "medcost": "$0.017370",
   "medtok": "20,914",
   "blended": "$0.977",
   "ratio": "16.3x"
  },
  {
   "cfg": "GPT-5.6 Luna",
   "iface": "a11y tree",
   "solved": "40/49",
   "medcost": "$0.019538",
   "medtok": "18,517",
   "blended": "$1.050",
   "ratio": "18.4x"
  },
  {
   "cfg": "Gemini 3.6 Flash",
   "iface": "computer use",
   "solved": "43/49",
   "medcost": "$0.020155",
   "medtok": "23,857",
   "blended": "$0.801",
   "ratio": "18.9x"
  },
  {
   "cfg": "GPT-5.6 Luna",
   "iface": "DOM + vision",
   "solved": "43/49",
   "medcost": "$0.032825",
   "medtok": "29,561",
   "blended": "$1.102",
   "ratio": "30.9x"
  },
  {
   "cfg": "Sonnet 5",
   "iface": "a11y tree",
   "solved": "42/49",
   "medcost": "$0.037782",
   "medtok": "10,762",
   "blended": "$3.255",
   "ratio": "35.5x"
  },
  {
   "cfg": "GPT-5.6 SOL",
   "iface": "computer use",
   "solved": "46/49",
   "medcost": "$0.062693",
   "medtok": "16,235",
   "blended": "$4.920",
   "ratio": "58.9x"
  },
  {
   "cfg": "Sonnet 5",
   "iface": "computer use",
   "solved": "39/49",
   "medcost": "$0.069705",
   "medtok": "57,701",
   "blended": "$1.365",
   "ratio": "65.5x"
  },
  {
   "cfg": "GPT-6 Astra",
   "iface": "code execution",
   "solved": "49/49",
   "medcost": "$0.118676",
   "medtok": "10,982",
   "blended": "$10.826",
   "ratio": "111.5x"
  },
  {
   "cfg": "Claude Opus 5",
   "iface": "computer use",
   "solved": "45/49",
   "medcost": "$0.138857",
   "medtok": "47,141",
   "blended": "$2.911",
   "ratio": "130.5x"
  },
  {
   "cfg": "Sonnet 5",
   "iface": "DOM + vision",
   "solved": "48/49",
   "medcost": "$0.210207",
   "medtok": "64,424",
   "blended": "$3.326",
   "ratio": "197.6x"
  },
  {
   "cfg": "GPT-6 Astra",
   "iface": "computer use",
   "solved": "45/49",
   "medcost": "$0.260780",
   "medtok": "20,560",
   "blended": "$12.855",
   "ratio": "245.1x"
  }
 ]
}