Live ranking, refreshed every 2 days
Best AI Models for Image Editing
As of October 6, 2026, the top-ranked model on LMArena's image edit leaderboard is gpt-image-2.5-sunburst by OpenAI with a score of 1,524, 43 points ahead of gpt-image-2.5-flare by OpenAI.
Updated October 14, 2026. Leaderboard data from October 6, 2026.
Ranking: top 10 models
Scores come from blind head-to-head votes on LMArena's image edit leaderboard. Rank is LMArena's published rank, which accounts for confidence intervals, so several models can share a rank. The top 10 are shown below, and ranks 11 to 40 follow in the expandable list.
| Rank | Model | Lab | Score | Votes | License |
|---|---|---|---|---|---|
| 1 | gpt-image-2.5-sunburst | OpenAI | 1,524 ±5 | 59,825 | Proprietary |
| 2 | gpt-image-2.5-flare | OpenAI | 1,481 ±5 | 57,408 | Proprietary |
| 3 | gpt-image-2 (medium) | OpenAI | 1,462 ±3 | 305,825 | Proprietary |
| 4 | grok-imagine-image-2.0 (canvas) | xAI | 1,439 ±9 | 11,589 | Proprietary |
| 5 | mai-image-2.6 | Microsoft Ai | 1,428 ±5 | 51,384 | Proprietary |
| 6 | gemini-nano-banana-2.1 | 1,428 ±6 | 12,985 | Proprietary | |
| 7 | grok-imagine-image-2.0 (20260801) | xAI | 1,425 ±5 | 50,427 | Proprietary |
| 8 | muse-image | Meta | 1,403 ±4 | 142,644 | Proprietary |
| 9 | mai-image-2.5 | Microsoft Ai | 1,401 ±4 | 190,081 | Proprietary |
| 10 | seedream-5.0-pro | ByteDance | 1,394 ±3 | 282,865 | Proprietary |
Show ranks 11 to 40
Rank Model Lab Score License 11 grok-imagine-image-quality (20260519) xAI 1,391 ±6 Proprietary 12 gemini-3-pro-image-2k (nano-banana-pro) Google 1,390 ±3 Proprietary 13 chatgpt-image-latest-high-fidelity (20251216) OpenAI 1,389 ±3 Proprietary 14 gemini-3.1-flash-image (nano-banana-2) [web-search] Google 1,387 ±3 Proprietary 15 gemini-3-pro-image-preview (nano-banana-pro) Google 1,386 ±3 Proprietary 16 reve-2.1 Reve 1,374 ±6 Proprietary 17 gpt-image-1.5-high-fidelity OpenAI 1,370 ±3 Proprietary 18 qwen-image-2.1 Alibaba (Qwen) 1,369 ±6 Open weights 19 reve-2.0 Reve 1,359 ±7 Proprietary 20 ideogram-4.5 Ideogram 1,347 ±5 Proprietary 21 uni-1.1-max Luma Ai 1,334 ±5 Proprietary 22 grok-imagine-image xAI 1,329 ±2 Proprietary 23 uni-1.1 Luma Ai 1,315 ±3 Proprietary 24 gemini-3.1-flash-lite-image (nano-banana-2-lite) Google 1,314 ±5 Proprietary 25 qwen-image-2.0-pro-2026-06-22 Alibaba (Qwen) 1,303 ±5 Proprietary 26 hunyuan-image-3.0-instruct Tencent 1,303 ±3 Open weights 27 wan2.7-image-pro Alibaba (Qwen) 1,303 ±4 Proprietary 28 seedream-4.5 ByteDance 1,302 ±2 Proprietary 29 wan2.7-image Alibaba (Qwen) 1,301 ±4 Proprietary 30 seedream-5.0-lite ByteDance 1,294 ±3 Proprietary 31 gemini-2.5-flash-image-preview (nano-banana) Google 1,293 ±2 Proprietary 32 seedream-4-2k ByteDance 1,270 ±7 Proprietary 33 reve-v1.1 Reve 1,262 ±3 Proprietary 34 flux-2-max Bfl 1,262 ±3 Proprietary 35 kling-image-o1 Kling 1,251 ±4 Proprietary 36 flux-2-pro Bfl 1,245 ±3 Proprietary 37 qwen-image-edit Alibaba (Qwen) 1,242 ±3 Open weights 38 reve-v1 Reve 1,235 ±5 Proprietary 39 qwen-image-edit-2511 Alibaba (Qwen) 1,234 ±3 Open weights 40 wan2.6-image Alibaba (Qwen) 1,231 ±3 Proprietary
Showing the top 40 of 59 models on this leaderboard. Data from October 6, 2026.
Best model by lab
Each lab's highest-ranked model on this leaderboard, and how far it sits from the top.
| Lab | Best model | Rank | Score | Behind #1 |
|---|---|---|---|---|
| OpenAI | gpt-image-2.5-sunburst | 1 | 1,524 | Leader |
| xAI | grok-imagine-image-2.0 (canvas) | 4 | 1,439 | 84 pts |
| Microsoft Ai | mai-image-2.6 | 5 | 1,428 | 95 pts |
| gemini-nano-banana-2.1 | 6 | 1,428 | 95 pts | |
| Meta | muse-image | 8 | 1,403 | 121 pts |
| ByteDance | seedream-5.0-pro | 10 | 1,394 | 130 pts |
| Reve | reve-2.1 | 16 | 1,374 | 149 pts |
| Alibaba (Qwen) | qwen-image-2.1 | 18 | 1,369 | 155 pts |
Which AI lab leads?
The lab with the most models at the top of the board is the one with the deepest bench, not only the single best model.
OpenAI holds 3 of the top 10 models (30%).
Models in the top 10 by lab
Show data table
| Lab | Top 10 |
|---|---|
| OpenAI | 3 |
| Microsoft Ai | 2 |
| xAI | 2 |
| ByteDance | 1 |
| 1 | |
| Meta | 1 |
Open vs proprietary: how big is the gap?
The best open-weight model is qwen-image-2.1 by Alibaba (Qwen) at 1,369, 155 points behind the best proprietary model, gpt-image-2.5-sunburst by OpenAI at 1,524. Their 95% confidence intervals do not overlap, so the gap is statistically meaningful.
17 proprietary models rank above the best open-weight model, and 1 of the top 20 is open-weight.
Open-weight here means any license other than Proprietary, including licenses with usage limits. It is a broader definition than the OSI one.
Other leaderboards
LMArena runs separate arenas for different jobs. The best model for one is often not the best for another.
Other arenas
- Text (overall)
- WebDev: claude-opus-5.5-max by Anthropic, 1,813
- Vision: claude-fable-5-high by Anthropic, 1,308
- Search: gpt-5.6-sol-xhigh by OpenAI, 1,256
- Document: claude-fable-5 by Anthropic, 1,499
- Text to image: gpt-image-2.5-sunburst by OpenAI, 1,425
- Text to video: gemini-omni-1.1-flash by Google, 1,516
Text arena by category
How to read this leaderboard
The image edit arena gives models an image and an instruction, such as removing an object or changing a style, and asks people to vote on the better edit.
- Scores come from blind votes
- LMArena shows a person two anonymous answers to the same prompt and asks which is better. Those head-to-head votes are fitted with a Bradley-Terry model, the statistical successor to Elo, to give each model one arena score. On the Elo scale LMArena uses, a 100-point lead means the higher model is expected to win about 64% of matchups.
- Confidence intervals decide ties
- Every score carries a 95% confidence interval. When two intervals overlap, the order of those two models is not settled. LMArena's rank column already accounts for this, which is why more than one model can hold the same rank.
- Why it differs from usage share
- A high rank does not mean many people use a model. Usage depends on price, distribution and habit. For adoption rather than quality, see AI market share.
Methodology and source
All rankings are read from the LMArena leaderboard dataset on Hugging Face, published by LMArena (formerly Chatbot Arena) under the Creative Commons Attribution 4.0 license. Credit for the underlying votes, scores and confidence intervals belongs to LMArena. Essentras is not affiliated with LMArena.
This page reads the image_edit subset, category overall, from the dataset's latest split. Scores, intervals, ranks, vote counts and license labels are shown as published. Essentras adds the grouping by lab and the derived figures on this page: gaps, lab shares, open vs proprietary and, on the main page, the history of who held #1.
Data refreshes automatically every 2 days. If the source cannot be reached, the page falls back to the last saved copy and says so at the top. Models are never merged or renamed; model and lab names appear exactly as LMArena publishes them.
Open-weight means any license other than Proprietary. History is limited to what the dataset records: the style-controlled text leaderboard from August 28, 2024. We do not scrape any other ranking site.
Track how AI models rank and describe your brand with our AI SEO service. Want these models to name your brand? Read our guide to LLM SEO.
Frequently asked questions
What is the best AI model for image editing?
As of October 6, 2026, gpt-image-2.5-sunburst by OpenAI ranks first on LMArena's image edit leaderboard with a score of 1,524. Rankings move as models and votes are added, so check the table above for the current order.
Which AI lab leads for image editing?
OpenAI holds 3 of the top 10 models (30%), ahead of Microsoft Ai (2), xAI (2), and ByteDance (1).
Is the best open-source model close to the best proprietary model?
The best open-weight model is qwen-image-2.1 by Alibaba (Qwen) at 1,369, 155 points behind the best proprietary model, gpt-image-2.5-sunburst by OpenAI at 1,524. 17 proprietary models rank above the best open-weight model, and 1 of the top 20 is open-weight. Their 95% confidence intervals do not overlap, so the gap is statistically meaningful. "Open" here means any license other than Proprietary, which is broader than the OSI definition.
What does this leaderboard measure?
The image edit arena gives models an image and an instruction, such as removing an object or changing a style, and asks people to vote on the better edit.
How often does this leaderboard update?
This page re-reads LMArena's dataset every 2 days. LMArena publishes a new snapshot when it has gathered enough votes or added models, so the date at the top shows the latest published data.
Models rank themselves. What do they say about you?
A leaderboard shows which AI models are strongest. It does not show what those models tell your customers about your company. Run the free check and see the answers.