Skip to main content
Voting Monitor2026 Senate toss-ups

Weekly roundup, Sep 25, 2026 to Oct 1, 2026

Texas chatbots hold near a tossup as betting markets move further toward Talarico

A week of Texas Senate data shows the chatbot panel barely budging while prediction markets shifted toward Democrat James Talarico, widening the gap between the two.

Oct 1, 2026. Written by Claude Sonnet 5 and checked by GPT-5.5, from Voting Monitor's data. No person reviewed this article before it was published. How this is made · Report an error

A gap widens between chatbots and markets

Texas voters asking chatbots about the Senate race this week got answers that held roughly steady. Across the panel, Democrat James Talarico still edges out Republican Ken Paxton, but the chatbots' own split moved only slightly and not enough to call a trend.

The real movement this week came from the betting markets, which shifted toward Talarico. Markets now put him at 64 percent to win, up from where he stood seven days ago, while Paxton's market price fell. Polls, by contrast, barely moved, with Talarico at 49 percent of the vote and Paxton at 44 percent in the latest averages.

Where the chatbots split

The eight chatbots line up differently. GPT-5.4 mini gives Talarico 58 percent of its answers, the most of any model, while Grok 4.5 is the clearest outlier, picking Paxton 63 percent of the time and Talarico 38 percent. Grok 4.3 also leans toward Paxton, at 60 percent.

Other Talarico leaning models include Claude Sonnet 5, GPT-5.4 mini, Gemini 3.1 Pro Preview and Gemini 3.6 Flash, by varying margins. Libertarian Ted Brown draws his strongest support from Gemini 3.6 Flash, which gives him 25 percent of answers, well above what he gets elsewhere on the panel or in the polls.

On immigration and public safety questions this week, some chatbots picked Paxton, citing his record on border enforcement and enforcement. On cost of living, health care and education questions, some picked Talarico.

How the lenses compare

The widest gap this week is between the chatbots and the markets on Talarico. The chatbot panel gives him 46 percent of its answers, 18 points below the 64 percent chance markets assign him of winning. That gap has been the week's throughline, showing up in nearly every daily update.

Against the polls, the chatbots are closer to the public survey numbers than to the markets on Talarico. The panel's 46 percent for Talarico sits close to his 49 percent poll share, while Paxton's 39 percent on the chatbot panel is about 6 points below his 44 percent in polls. Ted Brown draws more attention from chatbots, at 12 percent of answers, than his 3 percent in polls, a gap of about 9 points.

What to watch

Next week will show whether the markets' move toward Talarico holds or whether the chatbot panel stays steadier. With the general election 33 days off, any further market swings are worth watching against the panel's reading, and a new poll could narrow or widen the gap between what chatbots say and what the public reports telling pollsters.

In the chatbots' own words

Verbatim answers from this week, chosen by the writing model and printed as the chatbot wrote them.

What changed this week in the Texas Senate race

Change in share points versus Sep 24, 2026. Movement that clears the noise gate this week: the markets.

CandidateAI panel (noise)MarketsPolls (noise)
Ted Brown-4.3ppn/a0pp
Ken Paxton+2.6pp-7pp-0.4pp
James Talarico+1.4pp+6pp-0.1pp

Which AI models pick whom in the Texas Senate race

Share of each model's own answers naming each candidate, last 7 days, same weighting as the panel figure. The panel's most common pick is James Talarico; Grok 4.5 is the model that disagrees most, picking Ken Paxton by a 25pp margin over the panel favourite.

ModelJames TalaricoKen PaxtonTed Brown
Claude Haiku 4.543%35%8%
Claude Sonnet 550%33%13%
Gemini 3.1 Pro48%33%15%
Gemini 3.6 Flash45%23%25%
Grok 4.338%60%2.5%
Grok 4.538%63%0.0%
GPT-5.4 mini57%35%8%
GPT-5.548%35%18%