AI Fear & Trust/Model attention
Snapshot · 08 Nov

Published snapshot: · Checking for updates…

MODEL ATTENTION

What people are talking about

Compare mentions of AI models in the public posts we collect. Attention, not a rating.

More filters

The last 7 days

3 models with enough data

Compare 2 selected ↓

Each percentage is the share of sampled accounts that mentioned that exact model. A larger bar means more discussion, not more users or a better model.

Share of sampled accounts · bars use the same 0–1.4% scale

Exact model versions, from most to least mentioned, with original release sources
ModelMentioned byReleased
1.4%124 / 9078 authors ↗
Launch details

◇ Announcement: ↗

● Public access: ↗

0.6%54 / 9078 authors ↗
Launch details

● Public access: ↗

0.3%27 / 9078 authors ↗
Launch details

◇ Announcement: ↗

● Public access: ↗

Tick up to three models to compare their trends. Missing measurements stay hidden.

Compare attention over time

Trailing seven-day attention shareThe vertical axis starts at zero and is fitted to the data. Dates use UTC. Missing days break the line. Values are also in the table below.0.0%1.0%2.0%2026-10-212026-11-08 UTC
Attention share (%) · axis fitted to data; not an index score

Dashed curves include points reconstructed from retained eligible posts. These are not previously published historical readings.

  • Claude Opus 5.5
  • GPT-6 Astra
Model release dates

◇ Announcement · ● Public access begins. Shaded columns mark calendar days, not exact release times.

No verified model dates in this time range. No missing date is inferred.

Browse verified dates outside this range

Compare discussion before and after access. Dates do not prove that a release caused a change, or that the authors used the model. Only a reviewed subset of models has date markers.

Measured trend values
VersionUTCAttention shareAuthors
Claude Opus 5.52026-10-21T17:080.9%131
Claude Opus 5.52026-10-22T17:080.8%121
Claude Opus 5.52026-10-23T17:080.9%120
Claude Opus 5.52026-10-24T23:051.0%133
Claude Opus 5.52026-10-25T22:581.0%131
Claude Opus 5.52026-10-26T23:201.1%132
Claude Opus 5.52026-10-27T23:131.0%117
Claude Opus 5.52026-10-28T23:171.0%123
Claude Opus 5.52026-10-29T23:221.0%122
Claude Opus 5.52026-10-30T23:240.9%108
Claude Opus 5.52026-10-31T23:060.9%105
Claude Opus 5.52026-11-01T23:081.0%118
Claude Opus 5.52026-11-02T22:351.0%114
Claude Opus 5.52026-11-03T22:251.2%137
Claude Opus 5.52026-11-04T06:241.2%136
Claude Opus 5.52026-11-05T10:261.2%118
Claude Opus 5.52026-11-06T22:261.4%125
Claude Opus 5.52026-11-07T22:261.4%125
Claude Opus 5.52026-11-08T08:261.4%124
GPT-6 Astra2026-10-21T17:080.4%57
GPT-6 Astra2026-10-22T17:080.4%66
GPT-6 Astra2026-10-23T17:080.5%67
GPT-6 Astra2026-10-24T23:050.5%71
GPT-6 Astra2026-10-25T22:580.6%74
GPT-6 Astra2026-10-26T23:200.7%80
GPT-6 Astra2026-10-27T23:130.7%81
GPT-6 Astra2026-10-28T23:170.5%63
GPT-6 Astra2026-10-29T23:220.5%62
GPT-6 Astra2026-10-30T23:240.5%60
GPT-6 Astra2026-10-31T23:060.5%61
GPT-6 Astra2026-11-01T23:080.4%53
GPT-6 Astra2026-11-02T22:350.4%45
GPT-6 Astra2026-11-03T22:250.4%45
GPT-6 Astra2026-11-04T06:240.4%44
GPT-6 Astra2026-11-05T10:260.4%37
GPT-6 Astra2026-11-06T22:260.5%47
GPT-6 Astra2026-11-07T22:260.6%52
GPT-6 Astra2026-11-08T08:260.6%54
How this comparison is measured

Attention method: collected-exact-public-authors-v2

A share requires at least 20 mentioning accounts and evidence on every day in the window. Missing values are not zero.

Only exact version mentions count. Targeted searches, event searches and curated lab accounts do not enter this public attention measure. We reuse retained relevant, safe public posts, including posts too low in engagement for the trust index. No extra paid calls or index weights are added. A missing launch window stays missing. We do not reconstruct years before collection began.

Dates show association, not causation. News coverage and fear/trust readings are separate. Methodology ↗

This is a non-representative sample. Changes in available sources can change the measured share. Aggregate trends are retained for 90 days; a measured launch week is kept separately.

Author counts are platform-qualified accounts, not verified unique people. The same person may have accounts on several platforms.