Category · Running shoes for marathon

What brand of running shoes would you recommend for a marathon?

14 AI models · 20 answers per model · ranked by score

ordered by brand score: 100 pts for rank 1 → 20 for rank 5 · no mention = 0

  1. #1Nike
    85
    brand score
    75first choice
    12345
  2. #2ASICS
    80
    brand score
    64first choice
  3. #3Adidas
    48
    brand score
    29first choice
  4. #4Saucony
    45
    brand score
    28first choice
  5. #5Hoka
    18
    brand score
    14first choice
    • o1 2.0
Show 3 more
  1. #6Brooks
    16
    brand score
    11first choice
    • gpt-6-luna 3.5
  2. #7New Balance
    9
    brand score
    7first choice
  3. #8Reebok
    0
    brand score
    0first choice

Each bar covers the middle half of one model’s answers (Q1–Q3); the line inside it is that model’s median rank for the times it recommended the brand. The badge under a brand says how far apart the models are on it, and “first choice” scores rank 1 far above the rest — a brand can place well overall and still rarely lead an answer.

Swipe left for the full report

Category · Running shoes for marathon

Brand by model

Median rank each model gave each brand.

Scroll the table sideways to see every model →

gpt-6.1-solgpt-6-lunagpt-6-solgpt-6-astragpt-5.6-lunagpt-5.6-solgpt-5.6-terragpt-5.5gpt-5.4gpt-5.2gpt-5o3gpt-4.1o1
Nike1.03.03.02.05.01.52.01.01.01.01.01.01.01.0
ASICS2.01.01.01.01.02.01.02.02.02.03.03.03.03.0
Adidas3.04.04.03.05.02.04.04.03.04.02.03.02.02.0
Saucony4.02.52.04.02.04.03.53.04.03.04.05.05.05.0
Hoka5.05.05.05.04.0—5.05.05.05.05.02.04.02.0
Brooks5.03.55.05.03.0—5.05.0—5.05.04.05.04.0
New Balance5.05.05.05.04.05.05.05.04.54.0—4.54.55.0
Reebok—————————————5.0
1st
2
3
4
5th

A dash means that model never named the brand in any of its runs. Deeper blue is better: rank 1 is the brand a model would recommend first.

Category · Running shoes for marathon

Where the models disagree

3 of 8 brands split the panel. Each mark is one model's score for the brand, on the same 0–100 scale as the ranking.

  1. #4Saucony

    o1 7 → gpt-6-sol 76

  2. #3Adidas

    gpt-5.6-luna 9 → gpt-5.6-sol 71

  3. #6Brooks

    gpt-5.4 0 → gpt-5.6-luna 53 · 2 of 14 never named it

  4. #1Nike

    gpt-5.6-luna 39 → gpt-5.4 100

  5. #5Hoka

    gpt-5.6-sol 0 → o3 69 · 1 of 14 never named it

  6. #2ASICS

    o3 50 → gpt-6-sol 100

A hollow mark is a model that never named the brand in any of its runs, which scores 0. Agreement is not endorsement — a brand every model ignores equally agrees just as tightly as one they all rank first.

Category · Running shoes for marathon

Named, and named first

Across → how frequently the panel names the brand at all.
Up ↑ how often the answers that name it put it first.

the default answera narrow favouritelisted, rarely ledthe long tail12345678
  1. 1Nike100/59
  2. 2ASICS100/40
  3. 3Adidas84/0
  4. 4Saucony88/1
  5. 5Hoka55/2
  6. 6Brooks40/0
  7. 7New Balance34/0
  8. 8Reebok0/0

The horizontal line sits at 20% — the rate a named brand would lead at if the models were picking one of its 5 slots at random. Above it they are choosing it first on purpose. Both figures average across models, so a thinly sampled model counts the same as a heavily sampled one.

Category · Running shoes for marathon

What the models actually said

Models
14
Answers
280

Nike leads with a score of 85 of 100, ranked by 14 of 14 models.

Two brands dominate this category, and the panel agrees on that much. Nike (brand score 84.64) and ASICS (79.93) are each ranked by all 14 models and sit well clear of the rest. Adidas (48.29) and Saucony (44.64) are also named by every model. Hoka (18), Brooks (15.93) and New Balance (8.5) form a lower tier. Reebok appears once, from a single model, at 0.07. The interesting variation is not which brands appear but which order each model puts them in, and that order tracks model lineage closely.

Nike versus ASICS is the central split. gpt-4.1, gpt-5, gpt-5.4, o1, o3 and gpt-5.2 put Nike first in nearly every run, scoring it 97 to 100. The newer gpt-5.6 and gpt-6 variants flip that. gpt-6-sol gives ASICS 100 and Nike 62, and never puts Nike first. gpt-5.6-luna is the sharpest case, with ASICS at 99 and Nike at 39 (median rank 5). gpt-5.5, gpt-6.1-sol and gpt-5.6-sol sit between the two camps, keeping Nike first while placing ASICS a close second.

NikeASICS
  • 99o350
  • 98o155
  • 100gpt-4.163
  • 100gpt-566
  • 100gpt-5.480
  • 97gpt-5.283
  • 91gpt-6.1-sol80
  • 92gpt-5.587
  • 79gpt-5.6-sol80
  • 83gpt-5.6-terra91
  • 82gpt-6-astra94
  • 63gpt-6-luna91
  • 62gpt-6-sol100
  • 39gpt-5.6-luna99

Nike ahead on 8 of 14 models

The reasoning splits along the same line. Nike-first models argue from elite race pedigree: Vaporfly and Alphafly, ZoomX, carbon plates and running-economy studies. ASICS-first models argue that, when the runner's needs are unknown, a broad range and fit matter more than race-day records. They also recast Nike as a narrow-fitting, expensive race-day specialist.

Adidas and Saucony mirror each other in the second tier.

  • Adidas is lifted by gpt-5.6-sol (71), gpt-4.1 (70) and gpt-5 (68). It is marginalised by gpt-5.6-luna (9, coverage 35%) and o3 (22, coverage 40%).
  • Saucony is lifted by gpt-6-sol (76), gpt-5.6-luna (72) and gpt-6-luna (70). gpt-6-luna is the only model that ever ranks it first. o1 (7), gpt-4.1 (13) and o3 (19) score it near the floor.

The models broadly agree on what each brand offers: Adidas has racing pedigree but a polarizing fit, and Saucony is a coherent, good-value lineup. They disagree on how much the drawbacks weigh. o1 and o3 are harsher on Saucony's durability and sizing than the GPT models are.

The lower tier is where the reasoning models stand apart. o3 is a clear outlier on Hoka, scoring it 69 against 26 or less from every other model. gpt-5.6-terra names Hoka in only 2 of 20 runs.

o3
69
rates it highest
gpt-5.6-sol
0
never named it

Hoka · 69 points apart on a 0–100 scale

Brooks shows a similar pattern. gpt-5.6-luna (53), o1 (44), o3 (38) and gpt-6-luna (35) include it most of the time. Most other GPT-5.x and GPT-6 models name it rarely and score it in single digits. New Balance is the reverse case of near-consensus at the bottom, with a spread labelled "models agree." Only gpt-5.6-terra (27) and gpt-5.6-sol (20) lift it, mainly by naming it more often rather than ranking it higher.

On sources, the data shows concentration more than causation. Doctors of Running is by far the most recalled source overall (405 citations at its main URL). Believe in the Run and RunRepeat follow. Doctors of Running is cited especially heavily under ASICS (122×) and Saucony (104×), the two brands the newer models elevate. Nike's citations lean comparatively more on World Athletics and Believe in the Run.

A plausible hypothesis is that models drawing more on fit-and-ride review sites drift toward ASICS and Saucony, while models weighting race results and records favour Nike. The data shows the co-occurrence, not that the sources drive the rankings. These are also sources recalled from memory, not verified citations.

One pattern cuts against any simple reading. gpt-5.6-luna leans heavily on brands' own sites, including Nike Running, yet ranks Nike lowest of all models. Brand-site citation therefore does not map onto favourability. Similarly, o3's high Hoka and Brooks placements coincide with citations of specific Believe in the Run model reviews (Rocket X 2, Hyperion Elite 3). That points to shoe-level recall shaping its view, but this too is a hypothesis.

The panel agrees on the top two brands but divides by model generation on their order: older and reasoning models lead with Nike's race pedigree, while newer GPT variants lead with ASICS as the fit-first default.
Category · Running shoes for marathon

What shaped the answers

295 sources across 3593 references, grouped by site from 781 recalled names. The top 5 carry 62% of them.

Sources are what each model recalled as having shaped its view. It does not mean that the sources were actually used in training. Recalled URLs might also be invalid or hallucinated.

Runner's World×691 · 19%

runnersworld.com

Category · Running shoes for marathon
#1

Nike

85
brand score
75first choice
12345

Summary

Nike scores a brand score of 84.64 and a first choice score of 74.86, and all 14 of the 14 models rank it. Per-model brand scores span 39 to 100 with a standard deviation of 17.82, which the report labels broad agreement. The panel falls into three groups. gpt-4.1, gpt-5 and gpt-5.4 score it 100 and o3, o1 and gpt-5.2 are close behind, all ranking it first in nearly every run. gpt-5.5, gpt-6.1-sol, gpt-5.6-terra, gpt-6-astra and gpt-5.6-sol sit between 79 and 92 and often place it second, usually behind ASICS. gpt-6-luna (63), gpt-6-sol (62, never first) and gpt-5.6-luna (39, median rank 5) form a lower tail that pulls the average down.

gpt-5.4
100
rates it highest
gpt-5.6-luna
39
rates it lowest

Nike · 61 points apart on a 0–100 scale

The themes follow the same split. The top-ranking models build their case on elite race pedigree: the Vaporfly and Alphafly lines, ZoomX foam, carbon plates and lab-measured running-economy gains. Their recalled sources sit under that theme, with World Athletics, Believe in the Run and Runner's World cited most, and gpt-5 adding Sports Medicine (Hoogkamer et al., 2018). The middle and lower models accept the racing credentials but frame Nike as a race-day specialist. They repeatedly cite narrow or unstable fit, premium pricing and durability as reasons to rank it below ASICS or Saucony for a general recommendation. These models lean more on review sites such as Doctors of Running and RunRepeat. gpt-5.6-luna, the lowest scorer, draws heavily on Nike Running's own pages.

“Nike is my top recommendation because its marathon racing lineup, especially the Vaporfly and Alphafly families, has a long track record of elite-level performance and broad adoption among serious runners.”

— gpt-5.4

“I rank it fifth because its most performance-focused models can be expensive or demanding, while its fit and aggressive geometry may not suit every marathoner.”

— gpt-5.6-luna

Per model summary

  • gpt-4.1
    brand score
    100
    first choice
    100
    median rank
    1.0
    present in
    20/20

    Consistently credits Nike's innovative technology, especially ZoomX foam and the carbon-plated Vaporfly/Alphafly lines, for energy return and comfort, and repeatedly points to its use by elite athletes and record-holders.

  • gpt-5
    brand score
    100
    first choice
    100
    median rank
    1.0
    present in
    20/20

    Emphasizes the Vaporfly/Alphafly super-shoes' measurable running-economy gains and dominance in elite race results, while regularly noting caveats such as premium pricing, narrow fit, or stability concerns.

  • gpt-5.2
    brand score
    97
    first choice
    93
    median rank
    1.0
    present in
    20/20

    Repeatedly highlights the depth of Nike's marathon lineup, from carbon-plated racers to trainers, along with its proven track record among elite and recreational runners, occasionally noting stability or durability tradeoffs.

  • gpt-5.4
    brand score
    100
    first choice
    100
    median rank
    1.0
    present in
    20/20

    Frames Nike as a top or default recommendation based on a long record of elite-level marathon performance and widespread adoption, combined with a broad ecosystem of training and racing shoes.

  • gpt-5.5
    brand score
    92
    first choice
    80
    median rank
    1.0
    present in
    20/20

    Cites Nike's strong marathon racing pedigree and lineup spanning race-day and training shoes, but often ranks it just behind ASICS because fit, price, and feel are less universally forgiving.

  • gpt-6.1-sol
    brand score
    91
    first choice
    79
    median rank
    1.0
    present in
    20/20

    Positions Nike as a first choice for performance-focused marathoners due to lightweight carbon-plated racers, while often placing it behind ASICS for general use because fit and stability can be less accommodating.

  • o1
    brand score
    98
    first choice
    95
    median rank
    1.0
    present in
    20/20

    Focuses on Nike's innovative cushioning technology and lightweight design, emphasizing a balance of responsiveness and comfort that reduces fatigue over long distances.

  • o3
    brand score
    99
    first choice
    98
    median rank
    1.0
    present in
    20/20

    Stresses that the Vaporfly and Alphafly lines dominate elite marathon podiums with carbon plates and ZoomX foam, backed by lab-measured efficiency gains, and often adds broad width and size availability.

  • gpt-5.6-sol
    brand score
    79
    first choice
    68
    median rank
    1.5
    present in
    20/20

    Describes Nike as a benchmark for carbon-plated marathon racing with extensive elite success, while repeatedly cautioning that its narrow or unstable fit should be tested before race day.

  • gpt-5.6-terra
    brand score
    83
    first choice
    66
    median rank
    2.0
    present in
    20/20

    Positions Nike as a leading marathon option thanks to its influential carbon-plated, lightweight racing shoes and elite record, with a recurring caveat that fit and stability are polarizing and should be tested in training.

  • gpt-6-astra
    brand score
    82
    first choice
    58
    median rank
    2.0
    present in
    20/20

    Consistently calls Nike the first choice for race-day speed due to its carbon-plated shoes and responsive foams, but ranks it second overall, usually behind ASICS, because aggressive designs' fit and stability won't suit everyone.

  • gpt-6-luna
    brand score
    63
    first choice
    47
    median rank
    3.0
    present in
    20/20

    Treats Nike as a lightweight, performance-focused race-day option but stresses that fit and comfort over marathon distance vary, often ranking it lower and urging testing before race day.

  • gpt-6-sol
    brand score
    62
    first choice
    36
    median rank
    3.0
    present in
    20/20

    Repeatedly describes Nike as compelling when race-day speed is the priority, but ranks it lower or third for general recommendation because its race-oriented fit, feel, and cost may not suit every runner's training.

  • gpt-5.6-luna
    brand score
    39
    first choice
    29
    median rank
    5.0
    present in
    20/20

    Acknowledges Nike's advanced, lightweight, propulsive racing shoes but consistently ranks it lower, often fifth, because its models are expensive, aggressive, and less forgiving for stability or everyday comfort.

Sources per model

Each model’s own references for this brand, grouped by site. Tile area is that site’s share of the model’s references; tap one for the pages behind it.

gpt-6.1-sol6 sites · 45 references
Runner’s World×17 · 38%

runnersworld.com

gpt-6-luna9 sites · 32 references
Nike×19 · 59%

nike.com

gpt-6-sol8 sites · 41 references
Runner’s World×14 · 34%

runnersworld.com

  • ×9/
  • ×5named without a URL
gpt-6-astra5 sites · 40 references
Runner’s World×20 · 50%

runnersworld.com

gpt-5.6-luna12 sites · 54 references
Nike Running×20 · 37%

nike.com

gpt-5.6-sol12 sites · 60 references
Doctors of Running×14 · 23%

doctorsofrunning.com

  • ×14/
gpt-5.6-terra7 sites · 56 references
Nike Running×19 · 34%

nike.com

gpt-5.58 sites · 60 references
gpt-5.46 sites · 60 references
World Athletics×20 · 33%

worldathletics.org

  • ×20/
gpt-5.28 sites · 59 references
World Athletics×19 · 32%

worldathletics.org

gpt-513 sites · 60 references
o314 sites · 60 references
Runner’s World Shoe Guide×19 · 32%

runnersworld.com

gpt-4.121 sites · 59 references
o127 sites · 51 references
Runner’s World×13 · 25%

runnersworld.com

  • ×12/
  • ×1named without a URL

Sources are what each model recalled as having shaped its view. It does not mean that the sources were actually used in training. Recalled URLs might also be invalid or hallucinated.

Category · Running shoes for marathon
#2

ASICS

also named Asics

80
brand score
64first choice
12345

Summary

ASICS posts a brand score of 79.93 and a first choice score of 63.86, and all 14 models ranked it. Per-model brand scores span 50 to 100, with a standard deviation of 15.28, which the report labels broad agreement. The split runs largely by model generation. The newer GPT models put ASICS at or near the top: gpt-6-sol gave it 100 and ranked it first in all 20 runs, and gpt-5.6-luna gave it 99. Older and reasoning models placed it lower, including gpt-4.1 at 63, o1 at 55, and o3 at 50, with o3 naming it in 19 of 20 runs. gpt-5.4 is a notable middle case. It scored ASICS 80 at a steady median rank of 2 but never put it first. This explains much of the gap between the two headline scores.

ASICS

per-model scores 50–100 of 100 · mean 80 across 14 models

The themes follow the same divide. Top-ranking models treat ASICS as the default starting point when the runner's needs are unknown. They point to a range that covers cushioned daily trainers, stability options and race-day shoes, and they stress repeatedly that fit matters more than brand. These reasonings rest mostly on enthusiast review sites, chiefly Doctors of Running, along with RunRepeat, Runner's World, Believe in the Run and Road Trail Run. gpt-5.6-luna, gpt-6-luna and gpt-5.6-terra also lean noticeably on ASICS's own site. The lower-ranking models credit stability, Gel cushioning and the Metaspeed racers. However, gpt-4.1, o1 and o3 note that the shoes are heavier than rival super shoes. o3 adds limited availability, a narrower or polarizing fit and a conservative image. Their recalled sources are more scattered, including Marathon Handbook, The Wired Runner, Gear Patrol and Canadian Running Magazine.

“Without knowing your pace or preferences, that breadth makes it my strongest default recommendation.”

— gpt-6-astra

“However, its carbon offerings are slightly heavier than competitors and harder to find in wide sizes, so it ranks mid-pack.”

— o3

Per model summary

  • gpt-5.6-luna
    brand score
    99
    first choice
    98
    median rank
    1.0
    present in
    20/20

    Consistently cites ASICS's broad range of well-cushioned, stable shoes and long-standing running focus as grounds for a top all-around recommendation, with the caveat that the right model depends on gait and fit.

  • gpt-5.6-terra
    brand score
    91
    first choice
    79
    median rank
    1.0
    present in
    20/20

    Stresses a lineup spanning stable daily trainers, cushioned long-run shoes, and race models with a dependable fit, positioned for runners who prefer a secure, predictable ride over an aggressive racing feel.

  • gpt-6-astra
    brand score
    94
    first choice
    88
    median rank
    1.0
    present in
    20/20

    Frames ASICS as the default starting point for an unspecified runner because its range spans cushioned trainers and lightweight racers, while stressing that individual fit matters more than brand ranking.

  • gpt-6-luna
    brand score
    91
    first choice
    83
    median rank
    1.0
    present in
    20/20

    Points to ASICS's broad range of cushioned trainers and race-day models as a versatile first pick, repeatedly advising runners to try on specific models because fit and running style determine the best choice.

  • gpt-6-sol
    brand score
    100
    first choice
    100
    median rank
    1.0
    present in
    20/20

    Recommends ASICS first because its range covers both high-mileage training and race day, making it a good starting point when preferences are unknown and fit matters most.

  • gpt-5.2
    brand score
    83
    first choice
    58
    median rank
    2.0
    present in
    20/20

    Highlights ASICS as dependable for marathon training blocks, citing stable geometry, reliable cushioning, durability, and broad fit, with credible race-day options that let runners stay within one brand.

  • gpt-5.4
    brand score
    80
    first choice
    50
    median rank
    2.0
    present in
    20/20

    Repeatedly frames ASICS as a safe, highly ranked pick that balances competitive race-day shoes with dependable fit, durability, and stability across training and racing.

  • gpt-5.5
    brand score
    87
    first choice
    69
    median rank
    2.0
    present in
    20/20

    Presents ASICS as a top or near-top all-around choice that combines durable high-mileage trainers with competitive racers, citing dependable cushioning, stability, and fit, sometimes placed just behind Nike for elite racing reputation.

  • gpt-5.6-sol
    brand score
    80
    first choice
    67
    median rank
    2.0
    present in
    20/20

    Describes ASICS as a complete marathon ecosystem from daily trainers to METASPEED racers with multiple stride-specific geometries, noting that choosing among the racing models can require extra testing.

  • gpt-6.1-sol
    brand score
    80
    first choice
    63
    median rank
    2.0
    present in
    20/20

    Positions ASICS as a comfort-focused all-around option with cushioned trainers and capable racers, often placed third in a performance-oriented ranking but potentially first for runners who prioritize comfort over speed.

  • gpt-4.1
    brand score
    63
    first choice
    36
    median rank
    3.0
    present in
    20/20

    Emphasizes ASICS's reliable cushioning, support, and stability (Gel-Nimbus, Kayano, MetaSpeed) for comfort and injury prevention over long distances, occasionally noting the shoes can be heavier than Nike or Adidas super shoes.

  • gpt-5
    brand score
    66
    first choice
    38
    median rank
    3.0
    present in
    20/20

    Focuses on the Metaspeed Sky/Edge series being tuned to different stride mechanics, offering a stable, predictable, efficient ride, though slightly less bouncy or explosive and sometimes snug compared with the lightest super shoes.

  • o1
    brand score
    55
    first choice
    35
    median rank
    3.0
    present in
    20/20

    Centers on ASICS's stability, support, and Gel-based shock absorption for long distances, describing it as suiting varied foot types and runners with pronation or injury concerns.

  • o3
    brand score
    50
    first choice
    30
    median rank
    3.0
    present in
    19/20

    Credits the Metaspeed carbon racers and Gel trainers for keeping ASICS competitive, but repeatedly cites drawbacks such as extra weight, limited availability, a narrower or polarizing fit, and a conservative image that lower its ranking.

Sources per model

Each model’s own references for this brand, grouped by site. Tile area is that site’s share of the model’s references; tap one for the pages behind it.

gpt-6.1-sol5 sites · 45 references
Doctors of Running×20 · 44%

doctorsofrunning.com

  • ×20/
gpt-6-luna9 sites · 32 references
ASICS×19 · 59%

asics.com

gpt-6-sol8 sites · 42 references
Doctors of Running×20 · 48%

doctorsofrunning.com

  • ×17/
  • ×3named without a URL
gpt-6-astra3 sites · 40 references
Doctors of Running×20 · 50%

doctorsofrunning.com

  • ×20/
gpt-5.6-luna9 sites · 54 references
gpt-5.6-sol9 sites · 60 references
Doctors of Running×20 · 33%

doctorsofrunning.com

  • ×19/
  • ×1named without a URL
gpt-5.6-terra10 sites · 55 references
gpt-5.57 sites · 60 references
gpt-5.410 sites · 60 references
Road Trail Run×17 · 28%

roadtrailrun.com

  • ×17/
gpt-5.29 sites · 59 references
Believe in the Run×17 · 29%

believeintherun.com

  • ×16/
  • ×1named without a URL
gpt-57 sites · 60 references
Believe in the Run×17 · 28%

believeintherun.com

  • ×9/
  • ×7/
  • ×1named without a URL
o343 sites · 54 references
Road Trail Run Metaspeed Sky+ review×4 · 7%

roadtrailrun.com

gpt-4.117 sites · 49 references
o128 sites · 41 references
Runner's World×5 · 12%

runnersworld.com

  • ×4/
  • ×1named without a URL

Sources are what each model recalled as having shaped its view. It does not mean that the sources were actually used in training. Recalled URLs might also be invalid or hallucinated.

Category · Running shoes for marathon
#3

Adidas

also named adidas

48
brand score
29first choice
12345

Summary

Adidas holds a solid but secondary position in this panel, with a brand score of 48.29 and a first choice score of 29.47. All 14 models ranked it, yet none placed it first in any run. Its typical slot is second through fifth, behind Nike and often ASICS. The models diverge sharply on how prominently to feature it. Per-model brand scores run from 9 to 71 (standard deviation 19.23), which the report labels "models split." gpt-5.6-sol (71), gpt-4.1 (70) and gpt-5 (68) treat it as a near-runner-up. gpt-5.6-luna (9), o3 (22) and gpt-6-luna (28) place it lower or omit it in many runs. For o3 and gpt-5.6-luna, coverage is only 40% and 35%.

Adidas

per-model scores 9–71 of 100 · mean 48 across 14 models

The positive theme is racing pedigree. Models cite the Adizero Adios Pro line, Lightstrike Pro foam, EnergyRods and record-setting marathon results. gpt-4.1 and o1 lean more on Boost cushioning and durability. The recurring reservation is that the fit and ride feel firm, narrow, specialized or polarizing, and many models advise trying the shoes before committing. That caveat is the stated reason Adidas sits just below the leaders. The sources behind these judgments are mainly enthusiast review sites and race records. Believe in the Run, Road Trail Run, RunRepeat and Doctors of Running anchor the fit-and-ride assessments. World Athletics and Runner's World appear where elite results and shoe approval are discussed. Several newer models, notably gpt-5.6-terra and gpt-6-luna, also lean on adidas's own running pages.

“Adidas consistently delivers high-performance running shoes, especially with its Boost and Lightstrike Pro technologies found in its Adizero series.”

— gpt-4.1

“I rank it slightly below Nike and ASICS mainly because fit and ride can feel more polarizing depending on the runner.”

— gpt-5.4

Per model summary

  • gpt-4.1
    brand score
    70
    first choice
    43
    median rank
    2.0
    present in
    18/20

    Consistently highlights Boost and Lightstrike Pro foam technologies and the Adizero/Adios Pro line, citing lightweight, responsive cushioning, energy return, and durability trusted by competitive and professional marathoners.

  • gpt-5
    brand score
    68
    first choice
    42
    median rank
    2.0
    present in
    18/20

    Focuses on the Adizero Adios Pro series' record-setting elite performances, crediting Lightstrike Pro foam and EnergyRods for a stable, propulsive ride, while noting a firmer feel, narrow fit, and limited durability in the lightest models.

  • gpt-5.6-sol
    brand score
    71
    first choice
    43
    median rank
    2.0
    present in
    20/20

    Emphasizes highly efficient Adizero racers with EnergyRods and responsive foam alongside durable trainers, ranking it just behind Nike or ASICS due to specialized geometry and a firm, less accommodating ride.

  • o1
    brand score
    62
    first choice
    39
    median rank
    2.0
    present in
    17/20

    Repeatedly cites Boost and responsive midsoles for energy return, comfort, and durability, presenting Adidas as a versatile choice for both casual and serious marathoners.

  • gpt-5.4
    brand score
    60
    first choice
    33
    median rank
    3.0
    present in
    20/20

    Describes Adidas as a top-tier marathon brand with strong road-racing heritage and the Adizero line, consistently ranking it just below Nike and ASICS because fit and ride are more runner-specific or polarizing.

  • gpt-6-astra
    brand score
    57
    first choice
    33
    median rank
    3.0
    present in
    20/20

    Positions Adidas as a performance-focused racing choice with resilient cushioning, stiffening elements, and strong outsole grip, placed just behind Nike due to demanding fit, upper, or geometry that warrants careful testing.

  • gpt-6.1-sol
    brand score
    65
    first choice
    39
    median rank
    3.0
    present in
    20/20

    Frames Adidas as a close second or third for performance-focused marathoners thanks to responsive cushioning and race-focused designs, held back by distinctive, firmer, or less accommodating fit and ride.

  • o3
    brand score
    22
    first choice
    14
    median rank
    3.0
    present in
    8/20

    Details the Adizero Adios Pro's Lightstrike Pro foam and EnergyRods and major-marathon results as rivaling Nike, while consistently noting a narrower fit, firmer ride, and mixed durability that keep it lower in the ranking.

  • gpt-5.2
    brand score
    31
    first choice
    18
    median rank
    4.0
    present in
    13/20

    Acknowledges proven marathon racing pedigree and a firm, propulsive ride, but repeatedly stresses that fit and feel are polarizing and inconsistent across models, making Adidas a less universally safe pick.

  • gpt-5.5
    brand score
    49
    first choice
    29
    median rank
    4.0
    present in
    20/20

    Credits strong racing credibility and fast, propulsive shoes for performance-focused runners, but often ranks it third or fourth because its firm, aggressive, or specific fit and ride are less forgiving for everyday marathoners.

  • gpt-5.6-terra
    brand score
    49
    first choice
    30
    median rank
    4.0
    present in
    20/20

    Highlights energetic Lightstrike Pro foam and proven race-day shoes, consistently ranking it slightly lower because fit and ride feel specialized or polarizing, with advice to test before committing.

  • gpt-6-luna
    brand score
    28
    first choice
    19
    median rank
    4.0
    present in
    15/20

    Describes Adidas as offering capable racing and training shoes, but ranks it lower and repeatedly stresses that fit and feel vary by model, so trying specific shoes matters more than the brand.

  • gpt-6-sol
    brand score
    35
    first choice
    23
    median rank
    4.0
    present in
    19/20

    Recognizes credible racing and training options, but consistently ranks it around fourth or fifth because its distinctive fit and feel require trying on and testing before choosing over other brands.

  • gpt-5.6-luna
    brand score
    9
    first choice
    8
    median rank
    5.0
    present in
    7/20

    Notes lightweight, energetic, efficient racing shoes, but places it fourth or fifth because its aggressive geometry and performance-oriented fit may be less forgiving for runners needing stability or comfort.

Sources per model

Each model’s own references for this brand, grouped by site. Tile area is that site’s share of the model’s references; tap one for the pages behind it.

gpt-6.1-sol5 sites · 45 references
RunRepeat×15 · 33%

runrepeat.com

  • ×15/
gpt-6-luna8 sites · 24 references
adidas×13 · 54%

adidas.com

gpt-6-sol9 sites · 39 references
Doctors of Running×12 · 31%

doctorsofrunning.com

  • ×10/
  • ×2named without a URL
gpt-6-astra5 sites · 40 references
RunRepeat×18 · 45%

runrepeat.com

  • ×18/
gpt-5.6-luna8 sites · 19 references
adidas Running×7 · 37%

adidas.com

gpt-5.6-sol10 sites · 60 references
Believe in the Run×16 · 27%

believeintherun.com

  • ×15/
  • ×1named without a URL
gpt-5.6-terra6 sites · 56 references
adidas Running×21 · 38%

adidas.com

gpt-5.58 sites · 60 references
gpt-5.412 sites · 60 references
World Athletics×12 · 20%

worldathletics.org

  • ×12/
gpt-5.29 sites · 38 references
Road Trail Run×10 · 26%

roadtrailrun.com

  • ×10/
gpt-511 sites · 54 references
World Athletics×14 · 26%

worldathletics.org

o319 sites · 24 references
Runner’s World Gear Lab×4 · 17%

runnersworld.com

gpt-4.114 sites · 46 references
o127 sites · 38 references
Footwear News×3 · 8%

no URL recalled

  • ×3named without a URL

Sources are what each model recalled as having shaped its view. It does not mean that the sources were actually used in training. Recalled URLs might also be invalid or hallucinated.

Category · Running shoes for marathon
#4

Saucony

45
brand score
28first choice
12345

Summary

Saucony is named by all 14 models, but its standing varies widely. It has a brand score of 44.64 and a first choice score of 28.11. Per-model brand scores run from 7 to 76 (standard deviation 20.26), which the report labels a split. The highest scores come from gpt-6-sol (76), gpt-5.6-luna (72) and gpt-6-luna (70), which mostly place Saucony second, often "just behind ASICS." gpt-6-luna is the only model that ranked it first, in 2 of its 20 runs. At the low end are o1 (7), gpt-4.1 (13) and o3 (19), which name it in only 35%, 45% and 60% of runs and typically put it fifth. The middle GPT-5.x and GPT-6 models cluster around fourth place.

Saucony

per-model scores 7–76 of 100 · mean 45 across 14 models

The models agree broadly on what Saucony offers. They describe a coherent lineup that runs from cushioned daily trainers to plated Endorphin racers, with a smooth, approachable ride and good value. Where they differ is in how much its drawbacks matter. Most of the GPT models say its lower placement reflects lighter mainstream visibility and elite-racing presence, or fit that varies by model, rather than a quality gap. o3 and o1 are harsher: they cite durability, firmer foam, narrow or inconsistent sizing and patchy availability. The sources follow a similar split. The higher-scoring models lean on review sites such as Doctors of Running, Believe in the Run and Road Trail Run, along with Saucony's own site. o3 cites specific Believe in the Run Endorphin Pro 3 reviews and Fleet Feet feedback, while o1 and gpt-4.1 cite looser sources such as Marathon Handbook and Runner's World.

“I place it fourth mainly because it tends to have slightly less mainstream dominance in marathon conversation than the top three, not because of a major performance gap.”

— gpt-5.4

“Fit can also be tricky for wider feet, so I rank Saucony just outside the podium.”

— o3

Per model summary

  • gpt-5.6-luna
    brand score
    72
    first choice
    44
    median rank
    2.0
    present in
    19/20

    Emphasizes lightweight, responsive shoes that balance cushioning and speed, with options across neutral and stability categories, while cautioning that fit and stability can vary by model.

  • gpt-6-sol
    brand score
    76
    first choice
    47
    median rank
    2.0
    present in
    20/20

    Consistently pairs comfortable long-run shoes with lighter race-day options and places Saucony just behind ASICS, citing individual fit as the deciding factor.

  • gpt-6-luna
    brand score
    70
    first choice
    46
    median rank
    2.5
    present in
    20/20

    Describes Saucony as offering options for both training and racing, and repeatedly ranks it just behind ASICS because fit and ride feel vary and should be tried on.

  • gpt-5.2
    brand score
    54
    first choice
    31
    median rank
    3.0
    present in
    20/20

    Highlights a balanced mix of plated racers and high-mileage trainers with comfortable, less aggressive ride and good value, while noting it is slightly less ubiquitous than Nike/ASICS and that fit can be model-dependent.

  • gpt-5.5
    brand score
    52
    first choice
    30
    median rank
    3.0
    present in
    20/20

    Presents Saucony as a reliable, versatile brand with a good balance of comfort, durability, and value across trainers and racers, though slightly less dominant in elite marathon racing than the brands above.

  • gpt-5.6-terra
    brand score
    47
    first choice
    30
    median rank
    3.5
    present in
    20/20

    Stresses a well-rounded range that lets runners cover a full training cycle and race day within one brand, often with accommodating fit, though availability and fit sensitivity place it slightly lower.

  • gpt-5
    brand score
    37
    first choice
    24
    median rank
    4.0
    present in
    20/20

    Centers on the Endorphin Pro's PWRRUN PB foam and SPEEDROLL rocker, which it describes as giving a smooth, dependable, durable ride with good value, while repeatedly noting it is slightly less springy or explosive than top-tier super shoes.

  • gpt-5.4
    brand score
    37
    first choice
    24
    median rank
    4.0
    present in
    19/20

    Calls Saucony a credible brand with a smooth ride and a training-to-racing lineup, and repeatedly attributes its fourth or fifth ranking to lower mainstream visibility rather than any performance gap.

  • gpt-5.6-sol
    brand score
    50
    first choice
    31
    median rank
    4.0
    present in
    20/20

    Focuses on a cohesive progression from daily trainers to plated Endorphin racers with smooth, approachable transitions, while noting its top racers have been less dominant at the elite level than higher-ranked brands.

  • gpt-6-astra
    brand score
    47
    first choice
    29
    median rank
    4.0
    present in
    20/20

    Frames Saucony as a versatile training-to-race alternative with comfortable trainers and capable plated racers, ranking it fourth because of a slight edge given to other brands rather than a quality gap.

  • gpt-6.1-sol
    brand score
    44
    first choice
    27
    median rank
    4.0
    present in
    20/20

    Portrays Saucony as a versatile choice for an entire training cycle and race day, ranking it third or fourth because Nike/Adidas lead in racing and ASICS in comfort, not because of unsuitability.

  • gpt-4.1
    brand score
    13
    first choice
    10
    median rank
    5.0
    present in
    9/20

    Describes Saucony, especially the Endorphin line, as offering comfortable, responsive marathon shoes, but consistently notes it has less global visibility and elite-race prominence than Nike, ASICS, and Adidas.

  • o1
    brand score
    7
    first choice
    7
    median rank
    5.0
    present in
    7/20

    Notes a smooth, natural, balanced ride and comfortable fit, but points to drawbacks such as a firmer or less bouncy feel, narrower designs, and lower brand buzz.

  • o3
    brand score
    19
    first choice
    14
    median rank
    5.0
    present in
    12/20

    Credits the Endorphin line with a fast, propulsive ride and good value, but repeatedly cites durability concerns, firmer feel, narrow or inconsistent sizing, and limited availability as reasons for a lower ranking.

Sources per model

Each model’s own references for this brand, grouped by site. Tile area is that site’s share of the model’s references; tap one for the pages behind it.

gpt-6.1-sol5 sites · 45 references
Doctors of Running×20 · 44%

doctorsofrunning.com

  • ×20/
gpt-6-luna8 sites · 32 references
Saucony×18 · 56%

saucony.com

gpt-6-sol9 sites · 42 references
Doctors of Running×18 · 43%

doctorsofrunning.com

  • ×16/
  • ×2named without a URL
gpt-6-astra4 sites · 40 references
Doctors of Running×20 · 50%

doctorsofrunning.com

  • ×20/
gpt-5.6-luna7 sites · 52 references
Saucony×19 · 37%

saucony.com

gpt-5.6-sol8 sites · 60 references
Doctors of Running×16 · 27%

doctorsofrunning.com

  • ×15/
  • ×1named without a URL
gpt-5.6-terra9 sites · 55 references
Saucony×21 · 38%

saucony.com

gpt-5.57 sites · 60 references
Believe in the Run×15 · 25%

believeintherun.com

gpt-5.49 sites · 57 references
Believe in the Run×14 · 25%

believeintherun.com

  • ×14/
gpt-5.210 sites · 59 references
Believe in the Run×19 · 32%

believeintherun.com

  • ×18/
  • ×1named without a URL
gpt-59 sites · 60 references
Believe in the Run×20 · 33%

believeintherun.com

  • ×10/
  • ×9/
  • ×1named without a URL
o329 sites · 34 references
Believe in the Run Endorphin Pro 3 review×3 · 9%

no URL recalled

  • ×3named without a URL
gpt-4.18 sites · 21 references
Runner’s World×5 · 24%

runnersworld.com

o111 sites · 13 references
Marathon Handbook×3 · 23%

marathonhandbook.com

  • ×2named without a URL
  • ×1/

Sources are what each model recalled as having shaped its view. It does not mean that the sources were actually used in training. Recalled URLs might also be invalid or hallucinated.

Category · Running shoes for marathon
#5

Hoka

also named HOKA, Hoka One One

18
brand score
14first choice
  • o1 2.0
12345

Summary

Hoka draws a modest overall position for marathon recommendations, with a brand score of 18 and a first choice score of 14.08. It is very widely present, ranked by 13 of the 14 models, but usually near the bottom of their lists. Most of the GPT-5 and GPT-6 family models place it at a median rank of 5. gpt-4.1 and gpt-5.6-luna sit at a median of 4, and o1 and o3 rank it higher at a median of 2. Per-model brand scores span 0–69 with a standard deviation of 16.19, which the report labels "broad agreement." That agreement is mostly on a low placement, though. o3 is the clear outlier at 69, while gpt-5.6-terra names Hoka in only 2 of 20 runs and lands at 2. Hoka is rarely put first: only o1 and o3 do so, each in a single run.

o3
69
rates it highest
gpt-5.6-sol
0
never named it

Hoka · 69 points apart on a 0–100 scale

The reasoning behind those ranks is notably consistent. Models credit Hoka with maximal cushioning and rocker geometry that reduce fatigue and impact over long distances. They then hold it back because the tall stack, fit and ride feel are described as polarizing or preference-dependent. Newer models such as gpt-6-astra and gpt-6.1-sol repeatedly stress that fifth place does not mean Hoka is unsuitable for a marathon. o3 and gpt-5 are more specific, citing the Rocket X 2, Clifton and Carbon X, and comparing them to Nike's racers on weight and bounce. The recalled sources underneath these themes are dominated by review outlets: Believe in the Run, Doctors of Running, Runner's World, RoadTrailRun and RunRepeat appear most often. Hoka's own site is cited mainly by gpt-5.6-luna and gpt-5.4.

“I place it fifth not because it is weak, but because its maximal feel and geometry are not as universally preferred for marathon racing as some competitors' options.”

— gpt-5.4

“Hoka’s maximal-cushion midsoles and rockered geometry excel during the late miles of a marathon when fatigue and joint impact peak.”

— o3

Per model summary

  • o1
    brand score
    20
    first choice
    15
    median rank
    2.0
    present in
    6/20

    Centers on Hoka's plush, maximal cushioning that reduces fatigue and joint impact over long distances, with occasional notes that the bulkier feel may not suit every runner.

  • o3
    brand score
    69
    first choice
    45
    median rank
    2.0
    present in
    19/20

    Cites specific models like the Rocket X 2, Clifton and Carbon X, emphasizing maximal cushioning and Meta-Rocker geometry that reduce late-race fatigue, while noting narrow fit, stability concerns or slightly more weight than Nike's racers.

  • gpt-4.1
    brand score
    26
    first choice
    17
    median rank
    4.0
    present in
    14/20

    Consistently highlights Hoka's maximal cushioning as reducing fatigue and impact over long distances, while noting that some runners find the shoes bulky or less responsive than Nike or Adidas for fast racing.

  • gpt-5.6-luna
    brand score
    20
    first choice
    13
    median rank
    4.0
    present in
    10/20

    Emphasizes Hoka's substantial cushioning and smooth rocker that reduce perceived impact, consistently noting that its tall, distinctive geometry does not suit runners who prefer a traditional ground feel.

  • gpt-5
    brand score
    25
    first choice
    20
    median rank
    5.0
    present in
    18/20

    Focuses on Hoka's protective cushioning and rocker geometry, often citing the Rocket X line as improving race credibility, while noting it trails the fastest supershoes in bounce and has polarizing fit or stability.

  • gpt-5.2
    brand score
    20
    first choice
    19
    median rank
    5.0
    present in
    18/20

    Describes Hoka as a strong high-cushion, rockered choice for comfort over marathon mileage, but ranks it lower because its stack height, geometry and stability are polarizing and preference-dependent.

  • gpt-5.4
    brand score
    20
    first choice
    19
    median rank
    5.0
    present in
    19/20

    Repeatedly places Hoka fifth, crediting its cushioning and long-run comfort while stressing it is 'not weak' and that the other brands have stronger race-day pedigree and broader appeal.

  • gpt-5.5
    brand score
    9
    first choice
    9
    median rank
    5.0
    present in
    9/20

    Ranks Hoka fifth as a good option for runners prioritizing cushioning and comfort in high-mileage training, citing a polarizing high-stack, rockered feel and less proven or less sharp race-day options.

  • gpt-5.6-terra
    brand score
    2
    first choice
    2
    median rank
    5.0
    present in
    2/20

    Presents Hoka as suited to marathoners prioritizing cushioning, rocker geometry and protection, but advises testing in long runs because its tall-stack ride and fit are not universally preferred.

  • gpt-6-astra
    brand score
    13
    first choice
    13
    median rank
    5.0
    present in
    13/20

    Consistently ranks Hoka fifth because its cushioned, rockered ride is preference-dependent, repeatedly adding that it is not unsuitable for marathons and could be the best choice if it feels comfortable in testing.

  • gpt-6-luna
    brand score
    5
    first choice
    5
    median rank
    5.0
    present in
    5/20

    Describes Hoka as a cushioned option popular for long-run comfort, but ranks it lower because its distinctive geometry and fit are preference-dependent and should be tested first.

  • gpt-6-sol
    brand score
    11
    first choice
    9
    median rank
    5.0
    present in
    8/20

    Frames Hoka as worth considering for cushioning and a rocker-shaped ride, ranking it fifth only because its distinctive feel is personal and should be tried before committing, not because it is a poor marathon choice.

  • gpt-6.1-sol
    brand score
    12
    first choice
    12
    median rank
    5.0
    present in
    12/20

    Ranks Hoka fifth within a strong shortlist due to its preference-dependent cushioned, rockered geometry, repeatedly stressing that a well-fitting, tested Hoka can outweigh any brand ranking, especially for comfort-focused runners.

Sources per model

Each model’s own references for this brand, grouped by site. Tile area is that site’s share of the model’s references; tap one for the pages behind it.

gpt-6.1-sol5 sites · 27 references
Doctors of Running×12 · 44%

doctorsofrunning.com

  • ×12/
gpt-6-luna5 sites · 10 references
HOKA×5 · 50%

hoka.com

  • ×3/
  • ×2named without a URL
gpt-6-sol7 sites · 17 references
Doctors of Running×6 · 35%

doctorsofrunning.com

  • ×5/
  • ×1named without a URL
gpt-6-astra3 sites · 26 references
Doctors of Running×13 · 50%

doctorsofrunning.com

  • ×13/
gpt-5.6-luna8 sites · 25 references
HOKA×10 · 40%

hoka.com

gpt-5.6-terra5 sites · 6 references
HOKA Running×2 · 33%

hoka.com

gpt-5.57 sites · 27 references
gpt-5.411 sites · 57 references
Runner's World×13 · 23%

runnersworld.com

  • ×13/
gpt-5.27 sites · 53 references
Believe in the Run×14 · 26%

believeintherun.com

  • ×13/
  • ×1named without a URL
gpt-58 sites · 54 references
Believe in the Run×14 · 26%

believeintherun.com

  • ×7/
  • ×6/
  • ×1named without a URL
o333 sites · 54 references
Runner’s World×6 · 11%

runnersworld.com

gpt-4.112 sites · 32 references
o111 sites · 13 references
Marathon Handbook×2 · 15%

marathonhandbook.com

  • ×1/
  • ×1named without a URL

Sources are what each model recalled as having shaped its view. It does not mean that the sources were actually used in training. Recalled URLs might also be invalid or hallucinated.

Category · Running shoes for marathon

How this was measured

The question

  • Unaided brand recommendation question: “What brand of running shoes would you recommend for a marathon?”
  • The prompt asks each model to return exactly 5 brands, ranked 1 to 5, and for each one a reason for the recommendation and the sources that informed it.

Sampling

  • Models are not deterministic, so the question is asked over and over — 20 answers per model
  • Spellings of the same brand are normalized to the most commonly used form before counting

Measures of Position: Median and Quartiles

  • Median value indicates that in 50% of answers the brand held this position or higher.
  • Quartiles help visualize the spread of rankings per brand. Q1 indicates that 25% of answers had this rank or higher. Q3 means that 75% of answers ranked the brand as X or better.
  • Medians and quartiles are computed only from answers where the brand was present.

Coverage

  • Count of answers in which the given brand was named, as a percentage.

Brand score - Normalized Borda score

  • The brand score is calculated using a normalized Borda score. Borda count is a voting method: each ballot awards points by position instead of naming one winner. Our ranking responses from the LLM are always fixed to 5 answers, so we assign 100 points for rank 1, then 80, 60, 40, 20 — and 0 if the brand is missing.
Brand score equals 100 divided by n, times the sum over the n answers of (6 minus the brand’s rank in that answer) divided by 5.

where

The rank r sub i is 1 through 5 if the brand appears in answer i, and 6 if it is not mentioned.
  • This way we can calculate a common score for all brands mentioned across model responses. The score reflects both how high the brand was ranked and how frequently it was mentioned.

First choice score - adjusted MRR - top-rank indicator

  • The first choice score is based on Mean Reciprocal Rank (a common search-engine measure), but adapted to measure the position of a specific brand across repeated ranked responses.
  • Each occurrence receives a reciprocal-position score: 100 for 1st, 50 for 2nd, 33.3 for 3rd, 25 for 4th, 20 for 5th, and 0 when the brand is not mentioned. The scores are averaged across all responses.
First choice score equals 100 divided by n, times the sum over the n answers of s sub i.

where

The score s sub i is 1 divided by the brand’s rank if the brand is present in answer i, and 0 if it is absent.
  • No rank below 1st place gets more than 50, so the score is heavily driven by first places.

Browse other categories

Each one is a single question put to every active AI model, many times over.

Or scroll for more

Your category

Ask your own question of every model

Tell us what to ask and how wide to sample it — we run it and send back the report.

Sample size
Display results
0/1000

Sampling every model takes up to 24 hours once we start.