# DesignArena Image editing

> Source: Benchize (https://benchize.com) — https://benchize.com/benchmarks/designarena%3Aimage-to-image. When citing or reusing this content, credit "Benchize" and link to https://benchize.com/benchmarks/designarena%3Aimage-to-image.

Human pairwise preference evaluation of image editing. Voters compare model outputs for the same request; published Elo ratings and separate model configurations are retained for this specific arena and input modality.

Canonical URL: https://benchize.com/benchmarks/designarena%3Aimage-to-image
Last updated: Wed, 21 Oct 2026 18:52:20 GMT

Reproduction and quotation must name Benchize, link to the canonical HTML URL, and retain individual source citations.

Human pairwise preference evaluation of image editing. Voters compare model outputs for the same request; published Elo ratings and separate model configurations are retained for this specific arena and input modality.

Multimodal · Models tested: 9 · Score publishers: [designarena](https://www.designarena.ai/leaderboard/image-to-image)

## Model results

| # | Model | Score |  |
| --- | --- | --- | --- |
| 1 | [GPT Image 2.5 Sunburst](https://benchize.com/models/gpt-image-2-5-sunburst.md) | 1,404elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1404 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion gpt-image-2.5-sunburst variationName public ### Reported statistics averageGenerationTimeMs 33976 battles 8200 losses 2366 standardError Not reported winRate 71.2 wins 5834 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label GPT-Image-2.5 Sunburst Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/0` |  |  |  |
| 2 | [GPT Image 2.5 Flare](https://benchize.com/models/gpt-image-2-5-flare.md) | 1,373elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1373 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion gpt-image-2.5-flare variationName public ### Reported statistics averageGenerationTimeMs 22181 battles 8363 losses 2786 standardError Not reported winRate 66.7 wins 5577 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label GPT-Image-2.5 Flare Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/1` |  |  |  |
| 3 | [GPT Image 2](https://benchize.com/models/gpt-image-2.md) | 1,346elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1346 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion gpt-image-2 variationName public ### Reported statistics averageGenerationTimeMs 131138 battles 15879 losses 4331 standardError Not reported winRate 72.7 wins 11548 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label GPT Image 2 Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/2` |  |  |  |
| 4 | [ChatGPT Image](https://benchize.com/models/chatgpt-image.md) | 1,293elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1293 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion hazel variationName public ### Reported statistics averageGenerationTimeMs 59764 battles 50083 losses 14503 standardError Not reported winRate 71 wins 35580 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label ChatGPT Image Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/6` |  |  |  |
| 5 | [GPT Image 1.5](https://benchize.com/models/gpt-image-1-5.md) | 1,253elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1253 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion cherryblossom variationName public ### Reported statistics averageGenerationTimeMs 53636 battles 59576 losses 22299 standardError Not reported winRate 62.6 wins 37277 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label GPT-Image-1.5 Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/12` |  |  |  |
| 6 | [Nano Banana Pro (Gemini 3 Pro Image Preview)](https://benchize.com/models/gemini-3-pro-image-preview.md) | 1,245elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1245 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion gemini-3-pro-image-preview variationName public ### Reported statistics averageGenerationTimeMs 18989 battles 32936 losses 11075 standardError Not reported winRate 66.4 wins 21861 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label Gemini 3 Pro Image Preview Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/15` |  |  |  |
| 7 | [Gemini 3.1 Flash Lite Image](https://benchize.com/models/gemini-3-1-flash-lite-image.md) | 1,196elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1196 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion gemini-3.1-flash-lite-image variationName public ### Reported statistics averageGenerationTimeMs 9271 battles 30410 losses 14792 standardError Not reported winRate 51.4 wins 15618 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label Gemini 3.1 Flash Lite Image (Nano Banana 2 Lite) Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/20` |  |  |  |
| 8 | [Wan 2.7](https://benchize.com/models/wan-2-7.md) | 1,152elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1152 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion wan-v2.7-edit variationName public ### Reported statistics averageGenerationTimeMs 22241 battles 9886 losses 4987 standardError Not reported winRate 49.5 wins 4899 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label Wan 2.7 Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/26` |  |  |  |
| 9 | [GPT Image 1](https://benchize.com/models/gpt-image-1.md) | 1,106elo |  |
| ### Original source values \| Metric \| Value \| Unit \| Scope \| Direction \| \| --- \| --- \| --- \| --- \| --- \| \| `elo` \| 1106 \| elo \| image-to-image \| Higher is better \| ### Configuration arenaType models category imagetoimage inputModality Not reported sourceModelVersion gpt-image-1-edit variationName public ### Reported statistics averageGenerationTimeMs 69303 battles 27694 losses 14780 standardError Not reported winRate 46.6 wins 12914 ### Source [Original source](https://www.designarena.ai/api/leaderboard#board=image-to-image) Source designarena Original model label GPT-Image-1 Source version public Observed 2026-10-21T16:38:05.190Z Evidence SHA-256 `abf3d99140598dcb4ad29257e22198f27ce2f79576132aae8b8be4850de7f5b5` Source locator `/data/33` |  |  |  |

---
Cite as: DesignArena Image editing — Benchize, https://benchize.com/benchmarks/designarena%3Aimage-to-image (retrieved 2026-10-21)