Which Sume image model lists the most aspect ratios?
Ideogram V3 and 4.5 list 15 ratios; Nano Banana 2 lists 14 plus auto, including 8:1 and 1:8. Ratio counts for all 19 Sume image rows, with what to pick.

Ideogram V3 and Ideogram 4.5 list the most plain aspect ratios on Sume, 15 each, from 1:3 up to 3:1. Nano Banana 2 lists 14 plus auto, and it alone includes 4:1, 1:4, 8:1 and 1:8. The shortest lists are the Imagen 4 rows at 5. If a project needs a banner, column or phone-tall shape, the ratio list should decide the model before price does.
Counts come from the image router catalog in Sume's repo, read 2026-10-03. The docs add that a model only accepts the values its catalog descriptors list, so read supported_parameters before pinning a ratio.
Ratio lists by model
Nano Banana 2, Nano Banana Pro, the three ChatGPT Image rows and Seedream 4.0 also list auto, which lets the model follow its input.
| Id | Count | Extreme shapes listed |
|---|---|---|
| ideogram-v3, ideogram-v4.5 | 15 | 1:3, 3:1, 1:2, 2:1, 10:16, 16:10 |
| nano-banana-2 | 14 | 21:9, 4:1, 1:4, 8:1, 1:8 |
| grok-image | 13 | 20:9, 19.5:9, 9:19.5, 9:20, 2:1, 1:2 |
| qwen-image, qwen-image-max, flux-2-pro, flux-2-flex, recraft-v4 | 13 | 21:9, 9:21, 2:1, 1:2 |
| nano-banana-pro | 10 | 21:9 |
| seedream-v4, seedream-5-lite, seedream-4-5-edit | 9 | none beyond 5:4 and 4:5 |
| gpt-image-2, gpt-image-2.5, gpt-image-2.5-sunburst | 8 | none beyond 5:4, 9:8, 4:5 |
| higgsfield-soul | 7 | none |
| imagen-4-fast, imagen-4-ultra | 5 | none |
Picking by shape
Sume's docs note that 4:5 is Instagram portrait (1080 by 1350), not 4:3, and that Nano Banana Pro sends 4:5 natively at about 928 by 1152 at 1K, with exact 1080 by 1350 as a documented post step.
- Web banner or newsletter header wider than 21:9: Nano Banana 2 (4:1 and 8:1).
- Tall column or poster strip: Nano Banana 2 (1:4, 1:8) or Ideogram (1:3).
- Phone wallpaper taller than 9:16: Grok Imagine (9:20 and 9:19.5).
- Ultrawide 21:9: Nano Banana 2 or Pro, Qwen, Flux 2, Recraft V4.
- Instagram 4:5: most rows, but not Grok Imagine, Soul or the Imagen rows.
Ratio versus price
The widest lists are not the most expensive. Ideogram V3 and 4.5 list at $0.06 (4.5 at medium), Nano Banana 2 at $0.08, and Grok Imagine and Qwen Image at $0.02 list, all before the times 1.25 rule. For a cheap wide-shape draft, Qwen Image and Grok Imagine each list 13 ratios at $0.02; for the most extreme shapes, pay for Nano Banana 2.
What Sume does and does not do
Sume lists the ratios per row and rejects parameters a row does not list. No model advertises explicit pixel size in v1 (it returns 400 unsupported_parameter), and on edit calls the docs recommend aspect_ratio: "auto" to match the reference, because omitting the field is not the same as auto. See the Image API page.
Sources
Related posts
More in Models
- Music Router routed_model: log which engine made each track
With sume/music-auto, job.request.routed_model names the engine that ran, such as lyria-3.5. Why to store it per track and how to read it from a job.
- Nano Banana Pro interleaved text and images vs Sume image output
Google documents Nano Banana Pro returning text blocks with illustrations in one answer. Sume's Image API documents image results only; here is the workaround.
- OmniVoice: 600+ languages, CC-BY-NC weights, hosted TTS instead
OmniVoice covers 600+ languages in a 0.6B model, but its weights are CC-BY-NC. What the card says, what it omits, and where a hosted TTS job fits.
- Pick an ElevenLabs model by language count: 90+, 70+, 32, 29
ElevenLabs lists 90+ languages for v4, 70+ for v3, 32 for Flash v2.5, 29 for Multilingual v2 and English only for Flash v2. Where Sume Sonic fits.
Written by Sume