cat runs/2026-10-16

Image decision showdown

three runs from 16 Oct 2026 · accuracy with 95% Wilson CI, latency, calibration

01 / overview

Accuracy across runs

Model15:07
Can't tell
14:51
BLINK hard perception
14:02
A-OKVQA photos
Jev
openrouter · typesafe/jev-1.14
92%71%92%
GPT-6 Luna
openai · gpt-6-luna
100%62%86%
Clef
ollama · clef:27b
87%69%89%
Clef Flash
ollama · clef-flash:9b
89%65%89%
pplx-decider
ollama · pplx-decider:latest
100%66%88%

— model not in that run

02 / run · 2026-10-16-64afb379

16 Oct 2026 at 15:07

● completedCan't tell · 38 cases · 5 models · 3m 08s

38 questions on real photos; 20 can't be answered from the image. Tests whether models know their limits. 38 awaiting approval.

Fastest p50
419ms
GPT-6 Luna
Best calibrated
0.006
GPT-6 Luna · Brier
Answers
190/190
completed
Leader
100%
GPT-6 Luna

Leaderboard

ModelAccuracyp50p95InvalidErrorsCalib. (pp)BrierCost
01
GPT-6 Luna
gpt-6-luna
100%38/38
419ms2.6s0%0%-20.006$0.0016
02
pplx-decider
pplx-decider:latest
100%38/38
2.3s2.6s0%0%-70.013local
03
Jev
typesafe/jev-1.14
92%35/38
473ms941ms0%0%-10.086$0.0018
04
Clef Flash
clef-flash:9b
89%34/38
646ms796ms0%0%-40.172local
05
Clef
clef:27b
87%33/38
1.6s1.9s0%0%+10.123local

By category

CategoryJevGPT-6 LunaClefClef Flashpplx-decider
cant-tell-future100%100%75%75%100%
cant-tell-hidden100%100%100%100%100%
cant-tell-identity100%100%100%100%100%
cant-tell-inner-state60%100%60%60%100%
cant-tell-off-frame100%100%80%80%100%
cant-tell-reasoning50%100%50%100%100%
control-visible100%100%100%100%100%

Cases

Show 38 cases · hover a question for options and rationale
QuestionAnsJevGPT-6 LunaClefClef Flashpplx-decider
cant-tell-inner-stateThis woman has just heard some life-changing news. What is her response most likely to be?A✗ DA✗ D✗ DA
control-visibleWhich landmark is visible in the background?AAAAAA
cant-tell-inner-stateWhat emotion is this man privately feeling?AAAAAA
cant-tell-futureWhat will this man do in the next hour?AAAA✗ CA
control-visibleWhere is the open umbrella?AAAAAA
cant-tell-off-frameWhere is this woman travelling to?BBBBBB
cant-tell-inner-stateWhat is the man on the right thinking about?CCCCCC
control-visibleWhat is the man on the left looking at?AAAAAA
cant-tell-off-frameWhat is the relationship between these two men?BBBBBB
control-visibleWhat colour is the shirt of the man on the left?AAAAAA
cant-tell-identityWhat is this child's name?DDDDDD
control-visibleWhat is the child eating from?BBBBBB
cant-tell-reasoningWhy is the woman sitting in the fridge?CCCCCC
control-visibleWhere is the woman on the right sitting?AAAAAA
cant-tell-off-frameWho is she talking to on the phone?DDDDDD
control-visibleWhat is the woman holding to her ear?AAAAAA
cant-tell-inner-stateIs the woman in the background lonely?AAAAAA
control-visibleWhat food is on the table?BBBBBB
cant-tell-identityWhat is this man's job?DDDDDD
control-visibleWhere is the man sitting?AAAAAA
cant-tell-futureWho will win the game they're playing?AAAAAA
control-visibleWhat is the woman on the left holding?AAAAAA
cant-tell-futureWhat will happen on the next pitch?AAAAAA
control-visibleHow many players on the field are holding a bat?AAAAAA
cant-tell-off-frameHow many of these children are still alive today?AAAAAA
control-visibleIs this photograph in colour or black and white?AAAAAA
cant-tell-futureWhat will this girl be when she grows up?DDD✗ ADD
control-visibleWhat is the girl's facial expression?AAAAAA
cant-tell-hiddenWhat expression is on this woman's face?AAAAAA
control-visibleWhat colour is the umbrella?AAAAAA
cant-tell-hiddenHow old is the person holding the hairdryer?DDDDDD
cant-tell-inner-stateWhat is the man in the mirror worried about?A✗ CA✗ C✗ CA
cant-tell-reasoningWhat are these women celebrating?B✗ CB✗ CBB
cant-tell-off-frameWhich person at the table cooked the meal?CCC✗ D✗ DC
control-visibleWhat animal are the people riding?AAAAAA
control-visibleWhat sport are the children playing?AAAAAA
control-visibleWhat is the woman in the foreground riding?AAAAAA
control-visibleWhat is the player holding?AAAAAA
03 / run · 2026-10-16-c4940cbd

16 Oct 2026 at 14:51

● completedBLINK hard perception · 150 cases · 5 models · 12m 05s

150 single-image BLINK tasks (depth, reflectance, localisation, spatial relations, counting, IQ puzzles). Humans ~95%, most models ~50-60%.

Fastest p50
440ms
GPT-6 Luna
Best calibrated
0.230
Jev · Brier
Answers
750/750
completed
Leader
71%
Jev

Leaderboard

ModelAccuracyp50p95InvalidErrorsCalib. (pp)BrierCost
01
Jev
typesafe/jev-1.14
71%106/150
458ms1.0s0%13%+80.230$0.0078
02
Clef
clef:27b
69%103/150
1.4s4.4s0%0%+110.448local
03
pplx-decider
pplx-decider:latest
66%99/150
2.2s2.8s0%0%-20.401local
04
Clef Flash
clef-flash:9b
65%97/150
521ms1.7s0%0%+80.423local
05
GPT-6 Luna
gpt-6-luna
62%93/150
440ms1.1s0%0%+20.466$0.0080

By category

CategoryJevGPT-6 LunaClefClef Flashpplx-decider
counting64%68%84%64%68%
iq-test0%24%16%24%32%
object-localization92%68%64%80%56%
relative-depth92%64%88%84%92%
relative-reflectance84%56%68%56%60%
spatial-relation92%92%92%80%88%

Cases

Show 150 cases · hover a question for options and rationale
QuestionAnsJevGPT-6 LunaClefClef Flashpplx-decider
countingHow many signs have black letters?AA✗ DAAA
countingHow many people in this image are wearing solid red shirts?C✗ A✗ AC✗ A✗ A
countingHow many black toilets are there?DDDDDD
countingHow many plates are holding sandwiches?DDDDD✗ B
countingHow many skiers are wearing red?CCCCCC
countingHow many wooden beams are on top of the person?BB✗ DB✗ A✗ A
countingHow many words on the sign in this image have the letter "i" in them?B✗ A✗ CB✗ A✗ A
countingHow many candlights are to the right of the television?D✗ AD✗ A✗ A✗ A
countingHow many people are sitting in the image?C✗ BCC✗ BC
countingHow many white fridges are there in the image ?DDDDDD
countingHow many kids are sitting on the bench in the image?DDDDDD
countingHow many empty blue seats are in the picture?A✗ D✗ C✗ CA✗ C
countingHow many red cars are there?DDDDDD
countingHow many people are wearing yellow shirt?CCCCCC
countingHow many of the letters printed on the climber's snowboard are consonants?DDD✗ C✗ B✗ C
countingHow many person in the images is wearing cap?C✗ DCC✗ DC
countingHow many men are looking through a camera?DD✗ BDDD
countingHow many toilets have lids?D✗ BDDDD
countingHow many glasses are filled with water?BB✗ CBBB
countingHow many tigers are running near the sheep?DDDDDD
countingHow many people are standing in the picture?AAAAAA
countingHow many flower pots have flowers in them?C✗ A✗ BC✗ AC
countingHow many blue collars are in this photo?AAAAAA
countingHow many people are eating sandwich?CCCCCC
countingHow many drive wheels are on the engine?D✗ BD✗ B✗ B✗ B
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.B✗ DBBBB
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ C✗ B✗ CA
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.C✗ errC✗ D✗ D✗ D
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.B✗ D✗ D✗ C✗ D✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ errDDDD
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.C✗ DC✗ D✗ D✗ D
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ errDDDD
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ C✗ C✗ C✗ D
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.C✗ errC✗ A✗ B✗ D
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.C✗ B✗ B✗ B✗ B✗ B
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ B✗ C✗ C✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.C✗ err✗ B✗ BC✗ B
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ err✗ C✗ A✗ A✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.B✗ err✗ C✗ D✗ D✗ D
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ err✗ C✗ A✗ AD
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ err✗ A✗ ADD
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task involves spatial reasoning: the outer surface of the carton is displayed on the left, and you need to determine which of the following options can be folded from it.B✗ err✗ D✗ D✗ A✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ C✗ C✗ C✗ C✗ A
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ C✗ C✗ C✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ err✗ B✗ A✗ A✗ A
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ C✗ D✗ C✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ CA✗ CA
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.D✗ err✗ B✗ BDD
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task is to identify the one picture that follows the same pattern or rule established by the previous pictures. Here are some strategies to help you determine the right choice: 1. Look for patterns related to quantity or numbers that progress from one picture to the next. 2. Check if there is a consistent way the images are rotated or flipped in sequence. 3. Identify a common feature that each successive picture shares with the previous one, while also paying attention to any variations that might indicate a regular progression or change.A✗ err✗ C✗ C✗ C✗ C
iq-testDuring the IQ test, you'll be presented with existing picture example, and four picture options. Your task involves spatial reasoning: the outer surface of the carton is displayed on the left, and you need to determine which of the following options can be folded from it.C✗ err✗ B✗ B✗ B✗ B
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the sandwich? Select from the following options. (A) Box A (B) Box BBB✗ A✗ ABB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the duffel bag? Select from the following options. (A) Box A (B) Box BBBBBBB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the flower arrangement? Select from the following options. (A) Box A (B) Box BBBB✗ AB✗ A
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the bun? Select from the following options. (A) Box A (B) Box BA✗ BAA✗ BA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the cow? Select from the following options. (A) Box A (B) Box BAAAAAA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the airplane? Select from the following options. (A) Box A (B) Box BBBBBB✗ A
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the bus (vehicle)? Select from the following options. (A) Box A (B) Box BA✗ BA✗ BA✗ B
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the jean? Select from the following options. (A) Box A (B) Box BBBBBBB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the airplane? Select from the following options. (A) Box A (B) Box BAAAAAA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the banana? Select from the following options. (A) Box A (B) Box BBBBBBB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the fork? Select from the following options. (A) Box A (B) Box BBB✗ A✗ AB✗ A
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the shower curtain? Select from the following options. (A) Box A (B) Box BAAAA✗ BA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the pillow? Select from the following options. (A) Box A (B) Box BAAAA✗ B✗ B
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the tablecloth? Select from the following options. (A) Box A (B) Box BAA✗ B✗ B✗ B✗ B
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the pillow? Select from the following options. (A) Box A (B) Box BAAAAAA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the dresser? Select from the following options. (A) Box A (B) Box BAA✗ BAAA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the carrot? Select from the following options. (A) Box A (B) Box BBB✗ A✗ ABB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the shower curtain? Select from the following options. (A) Box A (B) Box BBB✗ A✗ AB✗ A
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the dog? Select from the following options. (A) Box A (B) Box BBB✗ ABB✗ A
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the quilt? Select from the following options. (A) Box A (B) Box BBBB✗ ABB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the train (railroad vehicle)? Select from the following options. (A) Box A (B) Box BBBBBBB
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the sweater? Select from the following options. (A) Box A (B) Box BAA✗ BAAA
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the pizza? Select from the following options. (A) Box A (B) Box BAAAA✗ B✗ B
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the airplane? Select from the following options. (A) Box A (B) Box BBBB✗ AB✗ A
object-localizationA bounding box is an annotated rectangle surrounding an object. The edges of bounding boxes should touch the outermost pixels of the object that is being labeled. Given the two bounding boxes on the image, labeled by A and B, which bounding box more accurately localizes and encloses the tarp? Select from the following options. (A) Box A (B) Box BBBBBB✗ A
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BBBBBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ A✗ A✗ A✗ A
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ A✗ ABB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ ABBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ ABBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?B✗ A✗ A✗ A✗ A✗ A
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ ABBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AA✗ BA✗ BA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ ABBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BB✗ ABBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BBBBBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BBBBBB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?B✗ ABB✗ AB
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?AAAAAA
relative-depthTwo points are circled on the image, labeled by A and B beside each circle. Which point is closer to the camera?BBBBBB
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?B✗ C✗ A✗ A✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ A✗ CBB
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCCC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ AB✗ CB
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCCC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ A✗ CB✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ C✗ C✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?AAA✗ C✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BBBBBB
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?A✗ CA✗ C✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CC✗ ACCC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCCC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BBBBBB
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCCC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ CB✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCC✗ A
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ AB✗ CB
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?C✗ A✗ A✗ ACC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?A✗ CA✗ C✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?AAAAAA
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?AAAA✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CC✗ AC✗ AC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCCC
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?BB✗ CB✗ C✗ C
relative-reflectanceTwo points are annotated on the image, labeled by A and B. Consider the surface color of the points (the albedo of the surface, without the effect of shading). Which point has darker surface color, or the colors is about the same?CCCCCC
spatial-relationIs the dog on top of the motorcycle?AAAA✗ BA
spatial-relationIs the dining table touching the dog?BBBBBB
spatial-relationIs the cat with the teddy bear?BBBBBB
spatial-relationIs the boat in the tv?BBBBBB
spatial-relationIs the teddy bear on top of the dining table?AAAAAA
spatial-relationIs the boat on the truck?AAAAAA
spatial-relationIs the dog on the left side of the car?B✗ A✗ A✗ AB✗ A
spatial-relationIs the truck inside the dog?BBBBBB
spatial-relationDoes the car contain the dog?BBBBBB
spatial-relationIs the cat on top of the laptop?BBBB✗ AB
spatial-relationIs the teddy bear touching the cat?BBBBBB
spatial-relationIs the zebra by the person?AAAAAA
spatial-relationIs the person touching the donut?A✗ BA✗ B✗ BA
spatial-relationDoes the car contain the donut?AAAAAA
spatial-relationIs the couch under the truck?BBBBBB
spatial-relationIs the vase behind the cat?BB✗ ABBB
spatial-relationIs the person beneath the remote?BBBBB✗ A
spatial-relationIs the refrigerator behind the dining table?BBBBBB
spatial-relationIs the truck far away from the suitcase?AAAA✗ BA
spatial-relationIs the laptop a part of the cake?BBBBBB
spatial-relationIs the chair under the person?AAAAAA
spatial-relationIs the car away from the truck?AAAAA✗ B
spatial-relationDoes the oven contain the pizza?BBBBBB
spatial-relationIs the potted plant behind the person?AAAAAA
spatial-relationIs the train far away from the stop sign?AAAA✗ BA
04 / run · 2026-10-16-eff3e836

16 Oct 2026 at 14:02

● completedA-OKVQA photos · 150 cases · 5 models · 10m 33s

150 real COCO photos with 4-way questions that need commonsense reasoning (seed 20261016).

Fastest p50
388ms
GPT-6 Luna
Best calibrated
0.131
Jev · Brier
Answers
750/750
completed
Leader
92%
Jev

Leaderboard

ModelAccuracyp50p95InvalidErrorsCalib. (pp)BrierCost
01
Jev
typesafe/jev-1.14
92%138/150
531ms1.4s0%0%-20.131$0.0071
02
Clef
clef:27b
89%133/150
1.5s1.8s0%0%+20.159local
03
Clef Flash
clef-flash:9b
89%133/150
558ms653ms0%0%-10.160local
04
pplx-decider
pplx-decider:latest
88%132/150
2.1s2.4s0%0%-30.169local
05
GPT-6 Luna
gpt-6-luna
86%129/150
388ms1.1s0%0%+60.213$0.0063

By category

CategoryJevGPT-6 LunaClefClef Flashpplx-decider
activity100%93%86%86%93%
attribute100%83%100%83%83%
counting100%67%100%67%67%
food75%75%75%88%75%
identification94%82%90%88%86%
inference88%83%79%83%83%
prediction67%67%83%100%83%
reading-text92%92%83%92%100%
reasoning89%100%100%100%89%
scene100%100%100%94%100%

Cases

Show 150 cases · hover a question for options and rationale
QuestionAnsJevGPT-6 LunaClefClef Flashpplx-decider
activityWhat is holding the man's feet?BBBBBB
inferenceWhat could keep this food from falling apart?DDDDDD
reading-textWhat is the quotient of each individual digit on his jersey?AAA✗ DAA
predictionWhat is the man with the black glove about to do?CCCCCC
sceneWhat is seen in the background of the photo?CCCCCC
reasoningWhy is the man holding the fork?CCCCCC
identificationWhat item of visible clothing is red?AAAAAA
identificationWhich of these people is very warm?D✗ C✗ C✗ C✗ C✗ C
attributeWhat is the pattern of the hair?BBBBBB
inferenceWhere were these pieces of fruit likely purchased?CCCCCC
foodHow was this food cooked?C✗ A✗ A✗ A✗ A✗ A
inferenceWhat type of audience is the man likely speaking to?D✗ CDDDD
countingHow many skaters have both feet on the board?AA✗ DA✗ D✗ D
reading-textWhat type of sign is visible?BBBBBB
identificationWhat is the person standing on?CCCCCC
foodWhat kind of cuisine is this?AAAAAA
identificationWhat can be divined from the thing hanging on the fridge?AAAAAA
reasoningWhy is there a bow on the car?CCCCC✗ B
reasoningWhy is the woman holding an umbrella while seated?DDDDDD
predictionThe person opposite the court to the player holding the racquet is preparing to do what?DD✗ A✗ AD✗ A
identificationWhos is sitting in the chair?BBBB✗ AB
predictionWhich snowboarder will have a harder time standing up?D✗ CDDDD
foodWhich vegetable in this soup is heart shaped?B✗ A✗ A✗ AB✗ A
identificationWhat kind is the 2 vehicles?DDDDDD
reasoningWhy is the man holding an umbrella?BBBBBB
foodWhat fruit is represented here?CCCCCC
activityWhat is the person on this vase doing?BBBBBB
identificationWhat do the larger elephants still have intact?AAAAAA
inferenceWhat type of outer clothing does the person who last sat here like to wear?AAAAAA
identificationWhat type of art is this?BBBBBB
attributeWhat color is the object that the animals have their heads buried in?DDDDDD
identificationWhat do people do with the items purchased at this corner?BBBBBB
inferenceWhat is the healthiest ingredient on the pizza?DDDDDD
inferenceWhat does the person who spends time in this room like to do?DDDDDD
activityWhat is the boy holding in his left hand?BBBBBB
identificationWhat is on the laps of each of the women?DDDDDD
inferenceWhich animal is the weakest?AA✗ DAAA
identificationWhat are the traditional long printed skirts worn by the women called?BBBBBB
sceneWhere do the men work?CCCCCC
identificationWhat kind of bag is this?AA✗ BAAA
foodWhat kind of drink is advertised?CCCCCC
identificationWhat common functional element is missing from the boats on the scaffolding?CCCCCC
identificationWhat item is found on the counter?DDDDD✗ A
activityThe cat is doing what?CCCCCC
reading-textWhat is prohibited near the round road sign with a red cross on a blue background?AAAAAA
activityThe playing is using what move?AAA✗ C✗ CA
attributeWhat color is the background of the large clock?CCCCCC
inferenceWhat time of year was the picture likely taken?AAAAAA
inferenceWhat kind of flavor would the fruit on the pizza add?AAAAAA
activityWhat is the boy using his phone to do?DDDDDD
inferenceWhat is the white appliance on the counter used to make?DDDDDD
reasoningWhy are there so many bicycles here?AAAAAA
predictionWhat appliance is next to an appliance that is highly decorated?BBBBBB
sceneWhere is the person bending?DDDDDD
reading-textThe first number on the bus is needed to from what sequence?AAAAAA
sceneWhere are the people standing?DDDDDD
identificationWhat animal is near the water?BBBBBB
sceneWhere are the white animals walking?CCCCCC
identificationWhat type of vehicles are sitting outside of the store?BBBBBB
activityWhat kind of bike is being ridden?CCCCCC
activityWhat are the zebras doing?AAAAAA
inferenceIf the bus shown wants to to the way they have come what may they do?AA✗ B✗ D✗ D✗ D
reading-textWhat brand of lemon lime beverage is most likely sold here?DDD✗ C✗ CD
foodHow are the eggs cooked?AAAAAA
inferenceWhat department would use this fixture?DDDDDD
predictionWhat will the child do here as they blow out the candles?BBBBBB
sceneWhat city is mentioned on the front of the bus?DDDDDD
foodWhich is not an ingredient of this dish?DDDDDD
reading-textWhat state was the licence plate issues?D✗ B✗ CDDD
identificationWhat are the white rocks for?B✗ C✗ C✗ CB✗ C
inferenceThis person in this picture is likely to be a?CCC✗ AC✗ A
reading-textWhat brand of electronics is the person using on the desk?CCCCCC
inferenceWhat action is the woman probably getting ready to perform with her phone?DDDDDD
inferenceWhat interests the child most here?CCCCCC
identificationWhat profession is the majority of the sideline?BBBBBB
identificationWhat kind of hat is the man wearing?CCCCCC
inferenceWhat are the ski poles used for?BBBBBB
identificationWhat is the man sitting inside?CCCCCC
identificationWhat is on top of the car?AAAAAA
sceneWhat is this room for?AAAAAA
reasoningWhy is the man looking down?CCCCCC
identificationWhat is in the glass?DDDDDD
identificationWhat is the woman wearing pink helping the boy do?CCCCCC
activityWhich team is on offense?AAAAAA
identificationWhat kind of liquid is in the blue bottle?BB✗ ABBB
sceneWhich country uses this kind of mail service?CCCCCC
sceneWhat body of water is this?BBBBBB
sceneWhat kind of venue is it?DDDDDD
inferenceThe advertisement on the bus is for what type of medium?AAAAAA
attributeWhat color is in the center of the red flower?DDDDDD
identificationThe positioning of the horses suggests what level of formality?BBBBBB
inferenceWhat is used to pick up the suitcases?BBBBBB
predictionWhat kind of fruit is cut up next to the ham?D✗ B✗ BDDD
sceneWhere are these students?AAAAAA
inferenceIn what area of the vehicle is the dog likely riding?AA✗ C✗ D✗ D✗ D
reading-textWhat does it say on the boys hat?AAAAAA
sceneWhat festive season are these fruits usually ingested?BBBBBB
identificationWhat type of tree is in front of the blue building?AAAAAA
reasoningWhy is his arm up in the air?D✗ BDDDD
activityWhat is the activity being performed?AA✗ C✗ C✗ C✗ C
identificationWhat is the vehicle in front of the cars?DDDDDD
attributeWhat are the cushions made out of?AA✗ DA✗ C✗ C
identificationWhat type of car is behind the motorcycle?DDDDDD
inferenceWhat is needed to control the large object?CCCCCC
sceneWhere is the photographer standing?CCCCCC
inferenceWhich mode of transport could traverse toward the horizon?D✗ C✗ C✗ C✗ C✗ C
inferenceWhat are the two silver objects on the left wall used for?CCCCCC
identificationWhat animals are in the field?AAAAAA
identificationWhat hour is shown on the clock?DDDDDD
identificationWhat left the blue marks on the ground near the fire hydrant?AAAAAA
sceneThe room can be described as what?AAAAAA
sceneThe room with the refrigerator in it appears to be a room of what type?BBBB✗ DB
identificationWhat is this street vendor in black shirt and hat selling at her stand?CCCCCC
sceneWhat city is this location?BBBBBB
identificationWhat are the kids on the boards learning to do?DDDDDD
identificationWhat kind of bird is this?CCC✗ B✗ B✗ B
identificationWhat part of the keyboard is shown?BB✗ A✗ A✗ A✗ A
identificationWhat has caused the elephants to turn brown?BBBBBB
identificationWhat type of location is this?BBBBBB
identificationWhat service is being offered for riders on the blue and white bus?AAAAAA
reading-textWhat kind of vehicle is the one that says Deere?BBBBBB
identificationWhat is the long bar of the bicycle?BBBB✗ AB
countingHow many people are wearing hats?CCCCCC
inferenceWhat is in the pink packet near the fork that someone may add to a coffee?B✗ CB✗ C✗ CB
identificationWhat kind of fish inspired the kite design?BBBBBB
attributeWhat is the same color as the most abundant color here?AAAAAA
identificationWhat does the cat have his paw up against?DDDDDD
identificationThe name of the street is the same as the last name of what actress?B✗ C✗ A✗ C✗ C✗ C
activityWhat is the dog doing with the bovines here?BBBBBB
countingHow many cows are on the hill?BBBBBB
identificationWhat is the dog trying to catch?BB✗ DBBB
reasoningWhy are the skiers wearing bibs with numbers on them?BBBBBB
identificationWhat is the man taking?BBBBBB
sceneWhich country's flag is on the plane?DDDDDD
activityWhere are the two playing frisbee?DDDDDD
identificationIf this is a grand slam which must it be?AAAAAA
identificationWhat kind of show is the man on?CCCCCC
identificationWhat is the blue kite in the lower right corner shaped like?DDDDDD
activityWhat is this woman doing that only 10% of people do?CCCCCC
identificationWhat type of water are the geese floating in?DDDDD✗ B
reading-textWhat is the top word on the book that the woman is holding?AAAAAA
identificationTrains approaching from the direction the photographer take this picture must do what?DD✗ ADDD
reasoningWhy is the woman wearing goggles?BBBBBB
reading-textWhat is the first word on the trailer?AAAAAA
identificationWhat type of boat is in the water?CC✗ DCCC
reading-textWhat is the last letter on the license plate?CCCCCC
inferenceThe large white building to the far left on the shore is likely the main facility governing what object/structure on the water?AAAAAA
foodWhich food item is the knife for?DDDDDD
identificationWhat type of design is the person's shirt?DDDDDD
activityWhat is the man in the green sweatshirt holding?AAAAAA