# Vision unable to read image\_url in model gpt-4-turbo, but it can in gpt-4-vision-preview?

**URL:** <https://community.openai.com/t/vision-unable-to-read-image-url-in-model-gpt-4-turbo-but-it-can-in-gpt-4-vision-preview/741952>\
**Category:** API\
**Tags:** chatgpt, gpt-4-vision, gpt-4-turbo\
**Created:** [May 10, 2024, 3:53pm UTC](https://community.openai.com/t/vision-unable-to-read-image-url-in-model-gpt-4-turbo-but-it-can-in-gpt-4-vision-preview/741952 "2024-05-10T15:53:17Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![rofungAI](https://avatars.discourse-cdn.com/v4/letter/r/8491ac/32.png) [@rofungAI](https://community.openai.com/u/rofungAI)\
**Post date:** [May 10, 2024, 3:53pm UTC](https://community.openai.com/t/vision-unable-to-read-image-url-in-model-gpt-4-turbo-but-it-can-in-gpt-4-vision-preview/741952/1 "2024-05-10T15:53:17Z")

</div>

```auto
{
"model": "gpt-4-turbo",
"max_tokens": "300",
"messages": 
	[
		{
			"role" : "user", 
			"content" : 
			[
				{
					"type" : "text",
					"text" : "what are the main colors in this image"
				},
       			{
      				"type" : "image_url",
      				"image_url" : 
      					{ 
      						"url" : "https://usvisabook.me/google-maps/clinic_images/Bangkok_Vadhana/040db5a6-fa9f-11ee-a39f-005056bf0cdf_8.jpg",
      						"detail" : "low"
      					}
      			}
      		]
    	}				

	]
}

```

I get response:

```auto
{
  "error": {
    "message": "Invalid image URL: 'messages[0].content[1].image_url.url'. Expected a base64-encoded data URL with an image MIME type (e.g. 'data:image/png;base64,aW1nIGJ5dGVzIGhlcmU='), but got a value without the 'data:' prefix.",
    "type": "invalid_request_error",
    "param": "messages[0].content[1].image_url.url",
    "code": "invalid_value"
  }
}

```

all other models return the same, until gpt-4-vision-preview and gpt-4-1106-vision-preview:

```auto
{
  "id": "chatcmpl-9NMnBARXtbQXz5cIGjO9slIXnCkQg",
  "object": "chat.completion",
  "created": 1715356185,
  "model": "gpt-4-1106-vision-preview",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The image appears to be an advertisement for a skin or beauty treatment service, possibly from a clinic or a beauty center. It features a woman with clear skin smiling and looking upwards, which suggests a sense of satisfaction or positive results after treatment. The imagery of what looks like a healthcare professional's hands performing a procedure on a client's skin might be indicative of the personalized care and services offered.\n\nThe text includes prices, suggesting that there is a promotional offer or a standard rate for the treatments. Unfortunately, I can't provide translations or personal opinions, but it is clear that the advertisement is targeted at people interested in skincare or cosmetic treatments, emphasizing affordability and professional service. There's also a date mentioned which could indicate a promotional period or a validity date for the offer. Additionally, a phone number is provided, presumably for potential clients to make appointments or inquire further about the services."
      },
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 95,
    "completion_tokens": 175,
    "total_tokens": 270
  },
  "system_fingerprint": null
}

```

I would almost use it as a workaround, but seems 4-vision-preview ignores any instructions and explains only whats in the image.

If I feed this image to chatGPT it’s much more detailed.

Anybody seeing this?

---

<div class="post-metadata">

**Author:** ![PaulBellow](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/paulbellow/32/597962_2.png) [@PaulBellow](https://community.openai.com/u/PaulBellow)\
**Post date:** [May 11, 2024, 3:39am UTC](https://community.openai.com/t/vision-unable-to-read-image-url-in-model-gpt-4-turbo-but-it-can-in-gpt-4-vision-preview/741952/3 "2024-05-11T03:39:28Z")

</div>

> [@rofungAI](#):
>
> explains only whats in the image.

What’s your… “text” : “#prompt#” … set as?

---

<div class="post-metadata">

**Author:** ![rofungAI](https://avatars.discourse-cdn.com/v4/letter/r/8491ac/32.png) [@rofungAI](https://community.openai.com/u/rofungAI)\
**Post date:** [May 13, 2024, 12:42am UTC](https://community.openai.com/t/vision-unable-to-read-image-url-in-model-gpt-4-turbo-but-it-can-in-gpt-4-vision-preview/741952/5 "2024-05-13T00:42:53Z")

</div>

i edited the “prompt” to be: “what are the main colors in this image” - still shows only the description of the image.

---

<div class="post-metadata">

**Author:** ![aivis](https://sea2.discourse-cdn.com/openai1/user_avatar/community.openai.com/aivis/32/370755_2.png) [@aivis](https://community.openai.com/u/aivis)\
**Post date:** [May 17, 2024, 7:53am UTC](https://community.openai.com/t/vision-unable-to-read-image-url-in-model-gpt-4-turbo-but-it-can-in-gpt-4-vision-preview/741952/6 "2024-05-17T07:53:41Z")

</div>

Try to change “max\_tokens”: “300” to “max\_tokens”: 300  
The max\_tokens value must be integer. Otherwise, this error comes up.