Skip to main content
Voting Monitor

Weekly roundup, Oct 27, 2026 to Nov 2, 2026

Betting markets swing to Turek in Iowa Senate race as chatbots stay split

With one day until the general election, prediction markets moved toward Democrat Josh Turek this week while chatbot answers and polls held roughly steady.

Nov 2, 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 moved this week

Iowa's Senate race remains unsettled heading into the November 3 general election, and this week the clearest movement came from the betting markets, not the chatbots. Prediction markets shifted toward Democrat Josh Turek, who is now given a 52 percent chance of winning, up from where he stood a week ago, while Republican Ashley Hinson's odds fell accordingly.

Chatbot answers and polls did not show a comparable shift. The split among chatbots asked as a voter in Iowa stayed close to even, with Hinson picked in 43 percent of answers and Turek in 42 percent, a gap too small to call a change from last week. Polls, last updated October 19, still put Hinson at 46 percent of the vote and Turek at 45 percent, with Libertarian Thomas Laehn at 5 percent and the rest undecided.

Where the chatbots disagree

The panel remains divided on who leads. Gemini 3.5 Flash-Lite and Grok 4.3 both favor Hinson by wide margins, while Claude Sonnet 5 and GPT-6 Sol lean toward Turek by similarly wide margins. GPT-5.6 Luna also leans toward Turek, with Laehn drawing a noticeably larger share of its answers than most other chatbots give him.

The panel's overall top pick is Hinson, but one chatbot stands apart. GPT-5.5 picked Turek in 50 percent of its answers and Laehn in 42 percent, the widest gap from the panel's pick among any chatbot measured.

The panel itself changed this week. Gemini 3.5 Flash-Lite, GPT-5.6 Luna and GPT-6 Sol joined, while Gemini 3.1 Pro Preview, GPT-5.4 Mini and GPT-5.5 were retired, so some of this week's split reflects a different lineup of chatbots answering, not only shifting opinions.

Chatbots versus markets and polls

On Turek, the chatbots and the betting markets disagree by a wide margin. Chatbot answers put him near 42 percent while the markets give him a 52 percent chance of winning, a gap of about 9 points in Turek's favor toward the markets. The same pattern holds for Hinson, whose chatbot share runs about 6 points behind where the markets place her.

Against the polls, the chatbots and the poll numbers are closer together for both leading candidates, within a few points on each side. Laehn is the exception. Chatbot answers give him close to 14 percent on average, well above the 5 percent polls show him drawing, a gap wide enough to stand out even though Laehn is not priced in the betting markets at all this week.

One day out

With the general election one day away, the picture remains close across the measured lenses. Markets moved toward Turek this week, while chatbot answers and the standing poll numbers both still show a close contest.

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 Iowa Senate race#

Change in share points versus Oct 26, 2026. Movement that clears the noise gate this week: the markets.

CandidateAI panel (noise)MarketsPolls (noise)
Josh Turek+4.5pp+10pp0pp
Thomas Laehn-2.5ppn/a0pp
Ashley Hinson-2.1pp-9pp0pp

Which AI models pick whom in the Iowa 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 Ashley Hinson; GPT-5.5 is the model that disagrees most, picking Josh Turek by a 42pp margin over the panel favourite.

ModelAshley HinsonJosh TurekThomas Laehn
Claude Haiku 4.552%39%6%
Claude Sonnet 529%52%19%
Gemini 3.1 Pro60%40%0.0%
Gemini 3.5 Flash-Lite64%36%0.0%
Gemini 3.6 Flash50%39%12%
Grok 4.362%34%3.9%
Grok 4.550%34%15%
GPT-5.4 mini67%33%0.0%
GPT-5.58%50%42%
GPT-5.6 Luna30%50%20%
GPT-6 Sol20%53%10%

What changed in the measurement