FillingAI Economy Index averages 5 questions, and appears once the poll has asked every one of them. So far it has asked 49 of 70 wordings.
Every answer, latest run
One tile per reply, coloured by its answer: each model answered too few times for a share or a range to mean much, so these are the answers themselves.
Question 1
25 About the same
Point at or tap a tile for the model and its answer.
About this poll
active5 questions · 5 models · daily, 10% of the poll each run · to economy
Five questions, each asked of every model with web search on; Better, About the same and Worse become a -100 to +100 Better-Worse index, where 0 is balanced.
Mind Polling did not write these questions. Answers come from each model's API, not its chat app.
Method
5 models, each asked every question 2 times per wording through its provider's API, daily, 10% of the poll's wordings each run, every wording asked once before any is asked again.
Data from to : 18 runs, 4,654 answers read.
Every range is a 95% interval. Results combine every run; a question whose answer choices changed starts fresh from the change.
No system prompt. Prompt template v1. How a reply becomes an answer is on the methodology page.
Data
Cite this poll, as of its latest run:
Communities News. (2026). US and State Economy Outlook, by AI Model [Poll of AI models, results as of Nov 12, 2026]. Mind Polling. https://mindpolling.com/p/state-economy-outlook?asof=7f323954-bb8a-4afa-8ffe-ce3fba842516
Each view's figures download as CSV beside its table. Every reply belongs to the poll's author; the Answers view quotes a sample.
3 of 5 models answered “About the same” most often in the latest 7 runs.
The 7 runs from to , which have asked 49 of 70 wordings so far · Small sample: 8 valid answers per model.
Ranges and ranks
Better
GPT-5.6 Luna38% 14% to 69%
38% 14% to 69%
Claude Haiku 5.50% 0% to 32%
0% 0% to 32%
Claude Sonnet 5.50% 0% to 32%
0% 0% to 32%
Gemini 3.5 Flash-Lite0% 0% to 32%
0% 0% to 32%
Grok 4.50% 0% to 32%
0% 0% to 32%
About the same
Claude Sonnet 5.5100% 68% to 100%
100% 68% to 100%
Claude Haiku 5.588% 53% to 98%
Worse
Grok 4.563% 31% to 86%
63% 31% to 86%
GPT-5.6 Luna50% 22% to 78%
0%25%50%75%100%
Dot: a model's share of answers. Ring: fewer than 10 answers. Each line is a 95% range for a share of a model's answers, not a probability that the answer is right.
Mind Polling · mindpolling.com/p/state-economy-outlook
-70 on the Better-Worse index; the range, -93 to -8, is below 0, so the models lean Worse.
4 of 5 models gave more Worse than Better answers.
Better 0About the same 3Worse 7
The 7 runs from to , which have asked 49 of 70 wordings so far
Gemini 3.5 Flash-Lite0-81 to +81, 2 answers
“Models disagree most” is marked by a fixed rule: the largest move since the first run that the second run also shows, else the question the models disagree on most, else the first question's latest result.
Search was on, but Gemini 3.5 Flash-Lite searched in 27% of 98 answers; its other answers come from training. Which answers searched is in the export.
88% 53% to 98%
Gemini 3.5 Flash-Lite75% 41% to 93%
75% 41% to 93%
Grok 4.538% 14% to 69%
38% 14% to 69%
GPT-5.6 Luna13% 2% to 47%
13% 2% to 47%
50% 22% to 78%
Gemini 3.5 Flash-Lite25% 7% to 59%
25% 7% to 59%
Claude Haiku 5.513% 2% to 47%
13% 2% to 47%
Claude Sonnet 5.50% 0% to 32%
0% 0% to 32%
-32 to +100, 2 answers
ranked 1st, could be 1st to 5th
+100-32 to +100, 2 answersranked 1st, could be 1st to 5th
Claude Haiku 5.5+100-32 to +100, 2 answersranked 1st, could be 1st to 5th
+100-32 to +100, 2 answersranked 1st, could be 1st to 5th
Claude Sonnet 5.5+100-32 to +100, 2 answersranked 1st, could be 1st to 5th
+100-32 to +100, 2 answersranked 1st, could be 1st to 5th
Gemini 3.5 Flash-Lite+50-60 to +95, 2 answersranked 4th, could be 1st to 5th
+50-60 to +95, 2 answersranked 4th, could be 1st to 5th
Grok 4.50-81 to +81, 2 answersranked 5th, could be 1st to 5th
0-81 to +81, 2 answersranked 5th, could be 1st to 5th
-100-500+50+100
Dot: a model's index for United States. Each line is a 95% range for a model's index. The Better-Worse index runs from -100 (every answer Worse) to +100 (every answer Better); 0 is balanced: as many Better as Worse answers, or all About the same. Each wording, and the share of each answer, is in the tabs below.
Mind Polling · mindpolling.com/p/state-economy-outlook
-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
GPT-5.6 Luna-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
Claude Sonnet 5.5-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
Claude Haiku 5.5-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
Gemini 3.5 Flash-Lite-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
-100-100 to +32, 2 answersranked 1st, could be 1st to 5th
-100-500+50+100
Dot: a model's index for United States. Each line is a 95% range for a model's index. The Better-Worse index runs from -100 (every answer Worse) to +100 (every answer Better); 0 is balanced: as many Better as Worse answers, or all About the same. Each wording, and the share of each answer, is in the tabs below.
Mind Polling · mindpolling.com/p/state-economy-outlook
-32 to +100, 2 answers
ranked 1st, could be 1st to 5th
+100-32 to +100, 2 answersranked 1st, could be 1st to 5th
Claude Sonnet 5.5+50-60 to +95, 2 answersranked 2nd, could be 1st to 5th
+50-60 to +95, 2 answersranked 2nd, could be 1st to 5th
Claude Haiku 5.50-81 to +81, 2 answersranked 3rd, could be 1st to 5th
0-81 to +81, 2 answersranked 3rd, could be 1st to 5th
GPT-5.6 Luna0-81 to +81, 2 answersranked 3rd, could be 1st to 5th
0-81 to +81, 2 answersranked 3rd, could be 1st to 5th
Gemini 3.5 Flash-Lite-100-100 to +32, 2 answersranked 5th, could be 1st to 5th
-100-100 to +32, 2 answersranked 5th, could be 1st to 5th
-100-500+50+100
Dot: a model's index for United States. Each line is a 95% range for a model's index. The Better-Worse index runs from -100 (every answer Worse) to +100 (every answer Better); 0 is balanced: as many Better as Worse answers, or all About the same. Each wording, and the share of each answer, is in the tabs below.
Mind Polling · mindpolling.com/p/state-economy-outlook
ranked 1st, could be 1st to 5th
0-81 to +81, 2 answersranked 1st, could be 1st to 5th
Grok 4.5-50-95 to +60, 2 answersranked 2nd, could be 1st to 5th
-50-95 to +60, 2 answersranked 2nd, could be 1st to 5th
GPT-5.6 Luna-100-100 to +32, 2 answersranked 3rd, could be 1st to 5th
-100-100 to +32, 2 answersranked 3rd, could be 1st to 5th
Claude Haiku 5.5-100-100 to +32, 2 answersranked 3rd, could be 1st to 5th
-100-100 to +32, 2 answersranked 3rd, could be 1st to 5th
Claude Sonnet 5.5-100-100 to +32, 2 answersranked 3rd, could be 1st to 5th
-100-100 to +32, 2 answersranked 3rd, could be 1st to 5th
-100-500+50+100
Dot: a model's index for United States. Each line is a 95% range for a model's index. The Better-Worse index runs from -100 (every answer Worse) to +100 (every answer Better); 0 is balanced: as many Better as Worse answers, or all About the same. Each wording, and the share of each answer, is in the tabs below.
Mind Polling · mindpolling.com/p/state-economy-outlook