What sizes does DALL·E 2 support?

DALL·E 2 generates square images at 256×256, 512×512 or 1024×1024 pixels. Unlike DALL·E 3, it does not offer landscape or portrait aspect ratios, so all outputs are square.

How much does each DALL·E 2 size cost?

At OpenAI's published rates, DALL·E 2 costs roughly $0.016 per 256×256 image, $0.018 per 512×512 image and $0.02 per 1024×1024 image. The size choice barely changes the price, so picking the largest you need is usually the better value.

Should I just always pick 1024×1024?

Often yes — at only a fraction of a cent more, 1024×1024 gives you the most resolution to crop, downscale or upscale from. Use smaller sizes only when you specifically need many tiny images cheaply and never plan to enlarge them.

Should I use DALL·E 2 or DALL·E 3?

DALL·E 3 follows prompts more faithfully and produces sharper, more coherent images, but costs more. DALL·E 2 is cheaper and supports editing and variations workflows. For final, high-quality assets prefer DALL·E 3; for cheap bulk drafts or variation experiments, DALL·E 2 can be fine.

What is the DALL·E 2 Size Picker?

Free DALL·E 2 size picker comparing the 256×256, 512×512 and 1024×1024 options with per-image cost, quality trade-offs and recommended use cases for thumbnails, social media and print, with a live batch total. It runs free in your browser on Gera Tools, with nothing uploaded.

DALL·E 2 Size Picker

Name: DALL·E 2 Size Picker
Creator: Gera Tools
License: https://creativecommons.org/licenses/by/4.0/

Get one useful tool a week

Like this tool? Enter your email and we'll send you one genuinely useful Gera tool a week — plus a link to come back to this one. No spam, one-click unsubscribe any time.

DALL·E 2 size picker

DALL·E 2 offers three square sizes — 256×256, 512×512 and 1024×1024 — and the price difference between them is tiny. This picker shows the per-image cost, quality trade-offs and best use case for each, and gives you a batch total so you can choose with full information before generating.

Comparing the three sizes

Size	Pixels	Best for	Relative cost
256×256	65,536	Icons, thumbnails, rapid prototyping	Lowest
512×512	262,144	In-feed social posts, app mockups	Slightly higher
1024×1024	1,048,576	Final assets, print source, upscaling	Highest (tiny margin)

The cost gap between the smallest and largest is only a fraction of a cent per image at OpenAI’s published rates, which is why the picker almost always recommends 1024×1024 unless you have a genuine reason to go smaller.

How it works

The tool maps each size to its published per-image rate and a recommended use. Because the cost gap between the smallest and largest is only a fraction of a cent, the recommendation usually favours generating at the largest size you might ever need and downscaling afterwards. Enter an image count to see the full batch total.

When smaller sizes do make sense

The main case for 256×256 is bulk disposable thumbnails — if you need hundreds of rough concept images for internal review and will never enlarge them, the lower per-image cost adds up. At 512×512 you get noticeably more detail than 256 while keeping a moderate file size, which suits in-feed social images where bandwidth matters. For anything you plan to use in production, 1024×1024 is the right choice.

DALL·E 2 vs DALL·E 3 — when to use which

DALL·E 2 is cheaper and supports two workflows DALL·E 3 does not: image editing (inpainting/outpainting) and image variations (generating riffs on an uploaded image). If your workflow involves editing existing images or exploring variations, DALL·E 2 is the appropriate model. For text-following accuracy and overall image quality on fresh generations, DALL·E 3 is clearly superior.

Tips

Default to 1024×1024. The extra resolution costs almost nothing and gives you room to crop and upscale from a single source.
DALL·E 2 is square-only. Unlike DALL·E 3 which has landscape and portrait options, every DALL·E 2 output is square — crop or add padding to fit non-square layouts.
Confirm current pricing. Rates may change; verify against your OpenAI billing dashboard before budgeting a large batch.
Variations and edits share the same size constraint. When using DALL·E 2’s edit and variation endpoints, the input and output image must match one of the three supported sizes.