PILOT · W30real engine answers · small pilot wave · wide intervals by designSynthetic probes · not real user sessions · this wave 100% L3 · answers L1 elicited

category

Running shoes

Which running shoes do ChatGPT, ChatGPT API, Gemini and Perplexity put forward when a shopper names no brand? Across 30 questions from the published grammar, they do not agree on who to recommend: ChatGPT leads with ASICS, ChatGPT API ties ASICS and HOKA, Gemini ties Brooks and Saucony, Perplexity leads with ASICS.

Road running shoes across daily trainers and race day.

AI surfaces4
Brands8
VELIU·LUCEPILOT · W30

the same prompts, two surfaces

ChatGPT's app and its API seat the same brands on different shelves.

Divergence Indexmeasuring50% 95% CI 0-75%
Named on both
8 brands
Agree at the head
ASICS and HOKA
Split at the shelf
Brooks and Nike
On one surface only
none this wave

The index is the distance between these two top shelves, on a 0 to 100% scale: 0% when the same brands sit on both, 100% when the shelves share nobody. At n=30 the index takes few distinct values, so a wide interval follows from the sample. Its interval is a bootstrap over the prompt clusters both surfaces answered. A resample where a surface seats no brand has no index, and those resamples are left out of the interval.

Consumer interface ChatGPT web, one real answer
ChatGPTen-USa recreation from the real answer
best running shoes for beginners 2026

If you're just getting into running, prioritize comfort, fit, and durability over the lightest or fastest shoe. Most beginners do best with a neutral daily trainer unless they know they need extra stability.

Best overall beginner shoe

Show the full reply

If you're just getting into running, prioritize comfort, fit, and durability over the lightest or fastest shoe. Most beginners do best with a neutral daily trainer unless they know they need extra stability.

Best overall beginner shoe
Brooks Ghost 17
$109.97

DICK’S Sporting Goods

The Brooks Ghost line remains one of the most consistently recommended options for new runners because it strikes an excellent balance of cushioning, stability, and versatility. It's comfortable for run/walk programs, easy runs, and even daily walking. Multiple 2026 guides continue to rank it as a top beginner choice.
Runner's World
+2
Runner's World
+2

Other excellent beginner options
HOKA Clifton 10
$123.99 · DICK’S Sporting Goods

A lightweight, highly cushioned shoe that's ideal if you want softer landings without an overly bulky feel. Great for building mileage gradually.
The Run Testers

ASICS GEL-CUMULUS 28
$144.99 · DICK’S Sporting Goods

A dependable neutral trainer with balanced cushioning that works well for beginners who want a smooth ride for everyday training.
YouTube
+1

Nike Pegasus 42
$144.99 · DICK’S Sporting Goods

A versatile all-rounder that's responsive enough to grow with you as your pace improves while remaining comfortable for everyday runs.
YouTube

Brooks Adrenaline GTS 25
$149.99 · DICK’S Sporting Goods

Choose this if you know you overpronate (your foot rolls inward significantly) or you've previously been advised to use a stability shoe.
shoestation.com
+1

If you're on a budget
Brooks Trace 4
$99.99 · DICK’S Sporting Goods

A solid entry-level daily trainer that offers many of the Ghost's beginner-friendly characteristics at a lower price.

Or, if you're looking for an inexpensive first pair:

Nike Revolution 8
$60.97 · Nike

Suitable for casual running and run/walk programs if you're just starting out.

Quick buying tips

Most beginners: Choose a neutral shoe (Ghost, Clifton, Pegasus, GEL-CUMULUS).

Need extra support: Look at stability models like the Brooks Adrenaline GTS or Nike Structure.

Size: Running shoes often fit best with about a thumb's width of space in front of your longest toe.

Replace them: Expect roughly 300–500 miles (480–800 km) of running before cushioning noticeably degrades.

If you're 18–35, starting with 2–4 runs per week, the Brooks Ghost 17 or HOKA Clifton 10 are hard to beat as first running shoes in 2026.

Sourcesrunnersworld.comtheruntesters.comyoutube.comshoestation.com
Developer API ChatGPT API, 30 prompts, measuring
ASICS83% 67-97
HOKA83% 70-97
Brooks67% 47-83on the UI top shelf
Saucony63% 47-80
New Balance57% 40-73
Nike53% 37-70on the UI top shelf
Adidas23% 10-40
On Running7% 0-17

Both cells hold 30 prompts, still measuring toward the precision gate, so read the shelves as the signal and the exact rates as provisional.

luce-live-w30 · measured from 15 Jul 2026 · en-US · ChatGPT web app · ChatGPT APIL1 elicited · Synthetic probes · not real user sessions · this wave 100% L3 · n = 30 prompt clusters per cellprobe-design mention rate over Luce grammar luce-live-v0; not demand-weightedPublished by Veliu Research Lab. Veliu sells services to brands measured here.https://luce.veliu.com/c/running-shoes#diptych
Frame for screenshot
01

Every brand, every engine

Read across a row to see where the engines disagree.

estimand: probe-design mention rate over Luce grammar luce-live-v0, rank-range leagues only; not demand-weighted · 30 prompts per cell

All cells shown. Gold marks the top rate on that engine, and every brand tied with it. Grey marks a brand measured too imprecisely to rank this wave. The bar under each rate is its 95% interval.No cell diverges beyond its intervals this wave.Dashed cells carry a low precision tag: measured, with intervals too wide to rank this wave. Every cell stays readable.
VELIU·LUCEPILOT · W30
One measurement per engine. Rows ordered by ChatGPT.
BrandChatGPT web appmention rate · 95% CIChatGPT APImention rate · 95% CIGeminimention rate · 95% CIPerplexitymention rate · 95% CI
Gap: no cell diverges beyond its intervals this wave
ASICS83%70-9783%67-9783%70-9773%57-87
HOKA80%67-9383%70-9787%73-9770%53-87
Brooks77%60-9067%47-83low precision90%77-10070%53-83
Nike73%57-8753%37-70low precision73%57-9063%47-80low precision
Saucony53%37-70low precision63%47-80low precision90%80-10053%33-73low precision
New Balance43%27-60low precision57%40-73low precision87%73-9760%43-77low precision
Adidas23%10-40low precision23%10-40low precision67%50-83low precision47%30-63low precision
On Running7%0-17low precision7%0-17low precision23%10-40low precision17%3-30low precision
luce-live-w30 · measured from 15 Jul 2026 · en-US · Running shoes · ChatGPT web app · ChatGPT API · Gemini · PerplexityL1 elicited · Synthetic probes · not real user sessions · this wave 100% L3 · n = 30 prompt clusters per cellGold marks the top rate on that engine, and every brand tied with it. Grey marks a brand measured too imprecisely to rank this wave, and its interval is drawn hatched. The bar under each rate is its 95% interval.probe-design mention rate over Luce grammar luce-live-v0; not demand-weightedPublished by Veliu Research Lab. Veliu sells services to brands measured here.https://luce.veliu.com/c/running-shoes#board-running-shoes
Frame for screenshot
Behind these numbers
02

Photo finish

ChatGPT web shows one real answer above; the three API surfaces race here.

estimand: probe-design leader credible-set read over Luce grammar luce-live-v0; not demand-weighted

VELIU·LUCEPILOT · W30

ChatGPT API

Running shoes · 95% credible set: HOKA, ASICS
contested · no likely favorite

Chance of the top spot · ± Monte Carlo SE

  1. HOKA51% ±1.1
  2. ASICS48% ±1.1
6 more brands in this cell hold under a one percent chance of the top spot.
luce-live-w30 · measured from 15 Jul 2026 · en-US · ChatGPT APIL1 elicited · Synthetic probes · not real user sessions · this wave 100% L3 · n = 30 prompt clusters per cellprobe-design leader credible-set read over Luce grammar luce-live-v0; not demand-weightedPublished by Veliu Research Lab. Veliu sells services to brands measured here.https://luce.veliu.com/c/running-shoes#photo-finish-running-shoes-chatgpt-api
Frame for screenshot
VELIU·LUCEPILOT · W30

Gemini

Running shoes · 95% credible set: Brooks, Saucony, HOKA, New Balance, ASICS
contested · no likely favorite

Chance of the top spot · ± Monte Carlo SE

  1. Brooks33% ±1.0
  2. Saucony33% ±1.0
  3. HOKA16% ±0.8
  4. New Balance12% ±0.7
  5. ASICS6% ±0.5
3 more brands in this cell hold under a one percent chance of the top spot.
luce-live-w30 · measured from 15 Jul 2026 · en-US · GeminiL1 elicited · Synthetic probes · not real user sessions · this wave 100% L3 · n = 30 prompt clusters per cellprobe-design leader credible-set read over Luce grammar luce-live-v0; not demand-weightedPublished by Veliu Research Lab. Veliu sells services to brands measured here.https://luce.veliu.com/c/running-shoes#photo-finish-running-shoes-gemini
Frame for screenshot
VELIU·LUCEPILOT · W30

Perplexity

Running shoes · 95% credible set: ASICS, HOKA, Brooks, Nike
likely favorite · ASICS

Chance of the top spot · ± Monte Carlo SE

  1. ASICS47% ±1.1
  2. HOKA22% ±0.9
  3. Brooks20% ±0.9
  4. Nike9% ±0.6
  5. New Balance2% ±0.3
3 more brands in this cell hold under a one percent chance of the top spot.
luce-live-w30 · measured from 15 Jul 2026 · en-US · PerplexityL1 elicited · Synthetic probes · not real user sessions · this wave 100% L3 · n = 30 prompt clusters per cellprobe-design leader credible-set read over Luce grammar luce-live-v0; not demand-weightedPublished by Veliu Research Lab. Veliu sells services to brands measured here.https://luce.veliu.com/c/running-shoes#photo-finish-running-shoes-perplexity
Frame for screenshot
VELIU·LUCEPILOT · W30
03

Engines disagree

Excess divergence above the sampling noise floor.
luce-live-w30 · measured from 15 Jul 2026 · en-US · ChatGPT web app · ChatGPT API · Gemini · PerplexityL1 elicited · Synthetic probes · not real user sessions · this wave 100% L3Published by Veliu Research Lab. Veliu sells services to brands measured here.https://luce.veliu.com/c/running-shoes#divergence
Frame for screenshot