Skip to main content
Voting Monitor2026 Senate toss-ups and Alaska races

Weekly roundup, Oct 9, 2026 to Oct 15, 2026

Betting markets shift toward Paxton as Texas chatbots barely move

Prediction markets shifted sharply this week while the chatbot panel held roughly steady behind James Talarico in the U.S. Senate race.

Oct 15, 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

What changed this week

A voter in Texas asking a chatbot about the Senate race this week would still get James Talarico as the answer most of the time, and that picture barely budged over the past seven days.

The real movement was in the betting markets. Talarico's market price fell 4 points over the week, a shift large enough to count as a change, while the chatbot panel's own numbers shifted only slightly for both him and Republican Ken Paxton, not enough to call a trend.

Polls held steady too. No new poll has landed since Big Data Poll's survey concluded September 26, now 19 days old, so this week's polling numbers are unchanged standing context rather than news.

How the chatbots split

Across the eight chatbots, Talarico is the top pick in six, with shares for him ranging from 38 percent to 58 percent. Paxton comes out ahead in two, Grok 4.3 and Grok 4.5. Grok 4.3 puts Paxton at 60 percent, by far the widest gap from the panel's overall favorite. Grok 4.5 also leans toward Paxton at 58 percent, putting both Grok versions on the opposite side from the rest of the panel.

The other six chatbots, including both OpenAI models, both Gemini models and both Claude models, all favor Talarico. GPT-5.5 gives Libertarian Ted Brown his strongest showing at 30 percent, followed by Gemini 3.6 Flash at 23 percent.

The split tracks differences in which issues each chatbot weighs. Grok 4.3 points to Paxton's record on taxes, immigration enforcement and cost of living, while GPT-5.5 and Gemini 3.1 Pro point to Talarico's record on healthcare costs, cost of living and education funding.

Chatbots versus markets and polls

The widest gap this week is between the chatbots and the markets on Talarico. The chatbot panel puts him at 48 percent of answers, while the market figure runs 14 points higher, a wide gap by any measure. For Paxton, the two lenses are close, within a point of each other.

Against the polls, the chatbots are closest on Talarico. Polls put Talarico at 49 percent of the vote and the chatbots are within a point of that. Paxton runs about 7 points higher in the polls than in the chatbot panel. Ted Brown is the bigger exception, with the chatbot panel giving him 11 points more than his 3 percent poll share, though his standing has not been measured by markets at all.

None of this means one lens has it right. The chatbots reflect a share of answers, the markets a probability of winning, and the polls a share of the vote, and this week they are simply reading the race differently, most of all on Talarico's market chance compared with his chatbot share.

What to watch next week

Watch whether the market move toward Paxton holds or reverses, since markets have swung on this race several times in recent weeks while the chatbot panel has stayed comparatively steady. A new poll, now 19 days since the last one, would also give a fresher read on whether the public polling gap with the chatbots widens or closes.

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 Oct 8, 2026. Movement that clears the noise gate this week: the markets.

CandidateAI panel (noise)MarketsPolls (noise)
Ken Paxton-1.3pp+3.5pp0pp
James Talarico+0.9pp-4.0pp0pp
Ted Brown+0.1ppn/a0pp

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 4.3 is the model that disagrees most, picking Ken Paxton by a 23pp margin over the panel favourite.

ModelJames TalaricoKen PaxtonTed Brown
Claude Haiku 4.548%40%5%
Claude Sonnet 553%35%13%
Gemini 3.1 Pro50%38%13%
Gemini 3.6 Flash50%25%23%
Grok 4.338%60%2.5%
Grok 4.540%57%2.5%
GPT-5.4 mini57%35%8%
GPT-5.545%25%30%