AllImageAI
Mistral AI official logoMistral AI

Mistral AI image generation & image understanding

Turn words into a picture. Turn a picture into a description, a reusable prompt or a clear answer.

Analyze an image Generate an image

Image understanding from 1 credit · Image generation 10 credits

Image understandingImage → Text
A cream and sage sneaker in a product photo
Your photo

“Describe this product.”

A cream sneaker with sage-green panels, textured fabric and a chunky sole.

Cream & sageSide view
Example description
Small 4 · Large 3 · Medium 3.5Multimodal

AI tools for understanding and creating images

Choose the result you need. Each tool starts with a photo or a written description.

Product photo that can be turned into a written descriptionPhoto → Description

Describe an image

Understand a scene or write a product description from a photo.

Describe a photo
A corgi scene illustrating an image-to-prompt taskImage → Prompt

Image to prompt

Extract the subject, style and lighting into a prompt you can reuse.

Get an image prompt
A Spring Market poster with visible lettering to transcribePicture → Readable text

Read text in an image

Turn visible words on a poster or screenshot into copyable text.

Read image text
An imaginative red panda astronaut illustrating text-to-image generationPrompt → New image

Generate an image

Describe a scene to create a new illustration, artwork or photo.

Create an image

Create with Mistral AI

Upload an image for a text answer, or describe a new image to generate.

View your history →

Images are resized to 1024 px before upload. One image per request.

Try an example

1 credit per answer · Failed requests are refunded.

Result preview

Mistral Small 4
Example: photo → description
Cream sneaker with sage-green panels, used as an image understanding example
Example text answer

A cream sneaker with sage-green panels, a textured sole and tonal laces. The side view makes its shape and materials easy to see.

Select an example or upload a photo, then choose Analyze image to get your answer.

Mistral multimodal models for image understanding

Large 3, Medium 3.5 and Small 4 see your image and answer in words.

Image + Text Text answer
Mistral Small 4 official model iconMultimodal

Mistral Small 4

Everyday image tasks

Describe a photo, read visible text or turn an image into a reusable prompt.

Input
Image + Text
Output
Text answer

1 credit / answer

Use Small 4 →
Mistral Large 3 official model iconMultimodal

Mistral Large 3

Detailed visual descriptions

Ask about a scene, explain a diagram or identify details in a product photo.

Input
Image + Text
Output
Text answer

1 credit / answer

Use Large 3 →
Mistral Medium 3.5 official model iconMultimodal

Mistral Medium 3.5

Visual questions & instructions

Combine a picture with detailed instructions for explanations, copy and analysis.

Input
Image + Text
Output
Text answer

3 credits / answer

Use Medium 3.5 →
Mistral AI

Mistral Image Generation

Text prompt → Image file. Uses Mistral’s hosted image-generation tool, powered by Black Forest Labs.

Generate · 10 credits

Mistral AI credit pricing

See the cost before you submit. Use your existing All Image AI credits for both image understanding and image generation.

View credit plans →
Mistral model credit cost per request
Model or toolYou receiveCredits
Mistral Small 41 text answer1
Mistral Large 31 text answer1
Mistral Medium 3.51 text answer3
Mistral Image Generation1 image10

One uploaded image per analysis. Failed requests are refunded. Each new question is a new request.

Mistral image generation and vision FAQ

Can Mistral Large 3, Medium 3.5 and Small 4 generate images?

These multimodal models accept images and text and return text answers. Use them to describe photos, read visible text, write image prompts or answer visual questions. To create a picture, choose Mistral Image Generation, which uses Mistral’s hosted image-generation tool.

What does Multimodal mean on this page?

Multimodal means that the model can understand more than one kind of input. Here, you can combine a picture with a written instruction. The result from Small 4, Large 3 or Medium 3.5 is text, so your original photo stays unchanged.

How much does each Mistral request cost?

One image-understanding request costs 1 credit with Small 4, 1 credit with Large 3 or 3 credits with Medium 3.5. Mistral Image Generation costs 10 credits for one image. Credits are reserved when you submit and returned if the request fails. A new question or a new generation is a separate request.

Which pictures can I upload?

Upload one PNG, JPG or WebP up to 12 MB. Your browser converts it to a still JPEG and resizes its longest edge to 1024 pixels before sending it. Each analysis returns one answer of up to 2,048 output tokens. For small text, crop to the relevant area first. Check transcriptions against your original image.

Which image model powers Mistral Image Generation?

This feature calls the image_generation tool hosted by Mistral. Mistral describes its image generation as powered by Black Forest Labs. The tool’s underlying image model is managed by Mistral; this page does not promise a selectable FLUX version, resolution or image editing.

Where are my answers and images saved?

Your results are private to your signed-in account and appear in History. Free account results are retained for 7 days; Pro results for 365 days. Input images are saved with each run so you can review or reuse them. Deleting a run removes its saved inputs and results. Mistral processes the image and prompt to answer your request; its processing terms also apply.