adjustable glasses illustration eye chart illustration clipboard illustration lenses illustration

Prompt: Unsealed box

46 out of 52 models answer this prompt correctly.
Prompt: Is the outer box sealed? Return yes or no.
Answer: No
Submitted by: Roboflow
Unsealed box

Contribute a Prompt

How models perform on this prompt

Arcee.ai Spotlight
ChatGPT-4o (Medium Reasoning)

Model answer: {"answer": "no"}

View more tests run on this model

Claude 3.5 Haiku

Model answer: No, the box appears to be partially opened or split.

View more tests run on this model

Claude 3.7 Sonnet

Model answer: No, the outer box does not appear to be sealed. The image shows a Kirkland Organic Coconut Water package that appears to be opened or at least not fully sealed, as you can see both sides of the packaging. The box is sitting on what looks like a wooden floor.

View more tests run on this model

Claude 4 Opus
Claude 4 Sonnet

Model answer: No, the outer box is not sealed. I can see that the box is open, revealing the product packaging inside.

View more tests run on this model

Claude 4.1 Opus
Claude 4.6 Opus

Model answer: {"answer": "no"}

View more tests run on this model

Cohere Aya Vision 32B
Cosmos Reason2 2B

Model answer: {"answer": "no"}

View more tests run on this model

Cosmos Reason2 8B

Model answer: {"answer": "no"}

View more tests run on this model

GLM 4.6v

Model answer: { "answer": "no" }

View more tests run on this model

GPT-4.1

Model answer: {"answer": "no"}

View more tests run on this model

GPT-4.1 Mini

Model answer: {"answer": "no"}

View more tests run on this model

GPT-4.1 Nano

Model answer: {"answer": "No"}

View more tests run on this model

GPT-5 Mini

Model answer: {"answer": "no"}

View more tests run on this model

GPT-5 Nano

Model answer: {"answer": "no"}

View more tests run on this model

GPT-5.4

Model answer: {"answer":"no"}

View more tests run on this model

Gemini 2.5 Flash

Model answer: {"answer": "no"}

View more tests run on this model

Gemini 2.5 Flash Lite

Model answer: {"answer": "no"}

View more tests run on this model

Gemini 2.5 Pro

Model answer: {"answer": "no"}

View more tests run on this model

Gemini 3 Flash

Model answer: {"answer": "no"}

View more tests run on this model

Gemini 3 Flash (Tools)

Model answer: {"answer": "no"}

View more tests run on this model

Gemini 3.1 Pro

Model answer: {"answer": "no"}

View more tests run on this model

Gemini 3.1 Pro (Tools)

Model answer: {"answer": "no"}

View more tests run on this model

Gemma 3 27b
Gemma 3 4B
Grok 4

Model answer: {"answer": "no"}

View more tests run on this model

Grok 4.1 Fast

Model answer: {"answer": "no"}

View more tests run on this model

Llama 4 Maverick 17B
Llama 4 Scout 17B
Mistral Small 3.1 24B
Molmo2 8B

Model answer: {"answer": "No"}

View more tests run on this model

OpenAI O1

Model answer: {"answer": "no"}

View more tests run on this model

OpenAI O4 Mini (Medium Reasoning)

Model answer: {"answer": "no"}

View more tests run on this model

Phi 4 Multimodal
Qwen 2.5 VL 7B
Qwen 3.5 122B (A10B)

Model answer: {"answer": "no"}

View more tests run on this model

Qwen 3.5 27B

Model answer: {"answer": "no"}

View more tests run on this model

Qwen 3.5 2B

Model answer: {"answer": "no"}

View more tests run on this model

Qwen 3.5 35B (A3B)

Model answer: {"answer": "no"}

View more tests run on this model

Qwen 3.5 4B

Model answer: {"answer": "no"}

View more tests run on this model

Qwen 3.5 9B

Model answer: {"answer": "no"}

View more tests run on this model

Qwen 3.5 Plus

Model answer: {"answer": "no"}

View more tests run on this model

Reka Edge

Model answer: { "answer": "no" }

View more tests run on this model

SmolVLM2 2.2B
Cohere Aya Vision 8B
Gemini 2.0 Flash

Model answer: {"answer": "yes"}

View more tests run on this model

Gemini 2.0 Flash Lite

Model answer: [ { "answer": "no" } ]

View more tests run on this model

Kimi k2.5

Model answer: {"answer": "yes"}

View more tests run on this model

Mistral Medium 3

Model answer: Yes. The outer box is sealed with a piece of tape across the top, as visible in the image.

View more tests run on this model

Qwen 3.5 0.8B

Model answer: { "answer": "yes" }

View more tests run on this model