Skip to content
ChangelogImproved

The Model Tracker comes to Gemini, Google AI Overviews and AI Mode

Model Tracker
The Model Tracker page with ChatGPT’s daily model share and the switch from gpt-5-6 to gpt-6, and the Google AI Overviews field list with each field’s share of answers

Follow the answering model, every response field and each day’s changes across ChatGPT, Gemini, Google AI Overviews and Google AI Mode, on one public page and a free API.

What changed

  • One tracker follows four engines: ChatGPT, Gemini, Google AI Overviews and Google AI Mode. Open querying.ai/en/model-tracker in eight languages and pick an engine at the top.
  • The eight component rates grow into every field of the response, at every depth: the share of answers that carry each field over the last 7 days, and the items per answer for list fields.
  • Fields are measured on three sides: the API response for every engine, the original stream for ChatGPT and Gemini, and the collector JSON for Google AI Overviews and Google AI Mode.
  • Each model switch lists the response fields whose share moved between the 7 days before and the 7 days after.
  • A change log dates every field that appeared, disappeared, changed its share or changed its items per answer, and selecting a field opens its daily history.
  • Three open endpoints under GET /v1/model-tracker return the same data as JSON. /chatgpt-model pages redirect to /model-tracker, and GET /v1/chatgpt-model answers 410 and points to the new endpoint.

How it works

The tracker counts every completed answer served through querying.ai for the four engines. Synthetic load tests and canaries are excluded. It counts field paths only; the text of questions and answers stays out of it. A field appears on the list once querying.ai’s own requests have carried it as well.

A share is the percentage of answers that carry a field, among the answers that carry its parent field. aioverview.shoppingCards.[].rating is read among the answers whose AI Overview shows shopping cards, so it holds steady on a day with fewer shopping questions. Items per answer is the average number of elements in a list field, among the answers that carry it. Current shares cover the last 7 days, and every rate rests on at least 200 answers.

API response is what querying.ai returns. Upstream is the original ChatGPT or Gemini stream, and collector JSON is the Google collector’s output for AI Overviews and AI Mode. Both are read before they are mapped to the API response, so you can follow a field from the engine’s output to the response you receive.

ChatGPT reports the model behind each answer, and the tracker shows the share of answers by model for the last 7 days and for each day since September 4, 2026. A switch is recorded when a different model leads three counted hours in a row. On October 9, 2026 at 07:00 UTC, ChatGPT moved from gpt-5-6 to gpt-6, and answers with entities went from 59.5% to 80.4%. Each switch compares the 7 days before it with the 7 days after the new model’s first day and lists the response fields that moved by at least one point. Read each move as a change over the same window: its causes include the model, a parser release and the mix of questions. Gemini and Google answers come without a model name, so those engines show fields and changes.

A change is logged when a field’s latest day leaves the range it held over the previous 7 days: the field appears, disappears, its share leaves that range by more than 15 points (or threefold and by at least 5 points for a small share), or its items per answer double or halve. The comparison starts once at least 3 of those days are measured, and a change enters the log when three hourly readings agree on its direction. Full field tracking for all four engines began on November 10, 2026, each field’s history starts on its first measured day, and every date is a UTC day.

Where it helps

  • On the day your brand’s mentions or citations move, check whether the engine switched models or a field such as sources or citationPills changed its share.
  • Before you build on a response field, read its share: a field at 100% arrives whenever its parent field does, and a field at 20% tells you how often to expect it.
  • Plan product, local and video content by engine: compare how often shopping cards, maps and videos arrive in ChatGPT, Gemini, Google AI Overviews and Google AI Mode.
  • Read the API once a day and alert your team when switches or changes gains an entry.

Get started

Call the API without a key. Responses are cached for 5 minutes, and each IP can send 240 requests per minute. Above that, the API answers 429 with RATE_LIMITED.

curl https://api.querying.ai/v1/model-tracker

Response

The response, trimmed for length.

JSON
{
  "success": true,
  "data": {
    "generatedAt": "…",
    "engines": [
      {
        "engine": "CHATGPT",
        "asOf": "2026-11-10",
        "models": { "gpt-6": 100, "gpt-6-mini": 0 },
        "modelDaily": [
          {
            "day": "2026-09-04",
            "shares": { "gpt-5-6": 96.8, "gpt-5-5": 3.2 }
          },
          "…",
          {
            "day": "2026-11-10",
            "shares": { "gpt-6": 100, "gpt-6-mini": 0 }
          }
        ],
        "switches": [
          {
            "at": "2026-10-09T07:00:00.000Z",
            "from": "gpt-5-6",
            "to": "gpt-6",
            "moved": [
              { "path": "citationPills", "before": 87.5, "after": 30.3 },
              { "path": "entities", "before": 59.5, "after": 80.4 },
              "…"
            ]
          }
        ],
        "fields": [
          "…",
          { "path": "entities", "share": 71.2, "perAnswer": 4.25 },
          { "path": "entities.[].name", "share": 100, "perAnswer": null },
          "…",
          { "path": "sources", "share": 96.2, "perAnswer": 18.3 },
          "…"
        ],
        "changes": []
      },
      "…",
      {
        "engine": "GOOGLE",
        "asOf": "2026-11-10",
        "models": {},
        "modelDaily": [],
        "switches": [],
        "fields": [
          "…",
          { "path": "aioverview", "share": 98.1, "perAnswer": null },
          {
            "path": "aioverview.citationPills",
            "share": 74,
            "perAnswer": 10.93
          },
          "…"
        ],
        "changes": []
      },
      "…"
    ]
  }
}
enginesarray
One entry per engine, named in engine: CHATGPT, GEMINI, GOOGLE for Google AI Overviews and AIMODE for Google AI Mode. asOf is the latest UTC day with answers.
engines[].modelsobject
The share of the last 7 days’ answers per reported model, in percent. ChatGPT reports a model name; Gemini and Google return {}.
engines[].modelDailyarray
Each day with at least 200 answers, oldest first, with shares per model in percent.
engines[].switchesarray
Every model switch with at, from and to. moved lists up to 12 response fields with their share in percent over the 7 days before and the 7 days after.
engines[].fieldsarray
Every field of the API response over the last 7 days. path is the field path, share the percentage of answers that carry it among the answers that carry its parent field, and perAnswer the average number of items in a list field (null where no list size is measured).
engines[].changesarray
Changes of the last 90 days, newest first, with at, side, path and kind: appeared, vanished, shifted for a share or resized for items per answer. from.low and from.high give the range of the previous 7 days, and to the new value.

Fields by side and by day

GET /v1/model-tracker/fields returns every field of one side of one engine over the last 7 days. GET /v1/model-tracker/field returns the daily history of one field in daily, oldest first, with one entry per day of at least 200 answers. Both take the query parameters below and answer 400 with INVALID_REQUEST when engine or side is missing.

curl -G https://api.querying.ai/v1/model-tracker/fields \
  -d engine=GOOGLE -d side=response

curl -G https://api.querying.ai/v1/model-tracker/field \
  -d engine=CHATGPT -d side=response -d path=entities
enginestringRequired
CHATGPT, GEMINI, GOOGLE for Google AI Overviews or AIMODE for Google AI Mode.
sidestringRequired
response for the API response, upstream for the original ChatGPT or Gemini stream, or collector for the collector JSON of Google AI Overviews and Google AI Mode.
pathstring
A field path as listed in fields, such as entities or sources.[].url. GET /v1/model-tracker/field requires it.
{
  "success": true,
  "data": {
    "engine": "GOOGLE",
    "side": "response",
    "fields": [
      { "path": "aioverview", "share": 98.1, "perAnswer": null },
      {
        "path": "aioverview.citationPills",
        "share": 74,
        "perAnswer": 10.93
      },
      "…",
      { "path": "organicResults", "share": 99.9, "perAnswer": 7.77 },
      "…"
    ]
  }
}

Pricing

The page and the API are free for everyone, with or without a querying.ai account.

More updates

Stay in the loop

Get new product updates in your inbox.

RSS feed