Image Edit Arena: LMArena Image Edit Arena
Crowdsourced human-preference leaderboard for image-editing models run by LMArena. Voters compare two anonymous edits of the same input image and instruction and pick the better one; ratings use a Bradley-Terry model with 95% confidence intervals. Higher is better.
Configurations
16
tracked on this benchmark
Score band
1524.0 – 1081.0
best → lowest tracked
Snapshot trend
—
need ≥2 snapshots
Leaderboard
Tracked models ranked by Arena Score (higher is better).
Configuration: Overall leaderboard entry (unqualified board entry)
Qualification: Preliminary (LMArena)
Notes: Arena score 1524 (95% CI 1518-1529), 59,825 votes, rank 1 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gpt-image-2.5-sunburst'. LMArena marks this score Preliminary. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Qualification: Preliminary (LMArena)
Notes: Arena score 1481 (95% CI 1476-1486), 57,408 votes, rank 2 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gpt-image-2.5-flare'. LMArena marks this score Preliminary. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Qualification: Preliminary (LMArena)
Notes: Arena score 1428 (95% CI 1422-1434), 12,985 votes, rank 6 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gemini-nano-banana-2.1'. LMArena marks this score Preliminary. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1401 (95% CI 1397-1405), 190,081 votes, rank 9 of 59, category overall, leaderboard published 2026-10-06. Board entry 'mai-image-2.5'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1386 (95% CI 1383-1389), 540,775 votes, rank 15 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gemini-3-pro-image-preview (nano-banana-pro)'. LLM Reference model is lifecycle-deprecated; score kept as historical. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1374 (95% CI 1368-1380), 20,792 votes, rank 16 of 59, category overall, leaderboard published 2026-10-06. Board entry 'reve-2.1'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Qualification: Preliminary (LMArena)
Notes: Arena score 1369 (95% CI 1363-1375), 21,941 votes, rank 18 of 59, category overall, leaderboard published 2026-10-06. Board entry 'qwen-image-2.1'. LMArena marks this score Preliminary. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1359 (95% CI 1352-1365), 19,265 votes, rank 19 of 59, category overall, leaderboard published 2026-10-06. Board entry 'reve-2.0'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1329 (95% CI 1327-1331), 756,681 votes, rank 22 of 59, category overall, leaderboard published 2026-10-06. Board entry 'grok-imagine-image'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1302 (95% CI 1300-1304), 1,294,217 votes, rank 28 of 59, category overall, leaderboard published 2026-10-06. Board entry 'seedream-4.5'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1242 (95% CI 1239-1244), 2,021,952 votes, rank 37 of 59, category overall, leaderboard published 2026-10-06. Board entry 'qwen-image-edit'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1231 (95% CI 1228-1233), 677,385 votes, rank 40 of 59, category overall, leaderboard published 2026-10-06. Board entry 'wan2.6-image'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1176 (95% CI 1174-1179), 6,404,567 votes, rank 52 of 59, category overall, leaderboard published 2026-10-06. Board entry 'flux-1-kontext-pro'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1139 (95% CI 1136-1141), 2,869,941 votes, rank 55 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gpt-image-1'. LLM Reference model is lifecycle-deprecated; score kept as historical. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1124 (95% CI 1121-1127), 682,914 votes, rank 56 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gpt-image-1-mini'. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Configuration: Overall leaderboard entry (unqualified board entry)
Notes: Arena score 1081 (95% CI 1079-1083), 4,942,834 votes, rank 57 of 59, category overall, leaderboard published 2026-10-06. Board entry 'gemini-2.0-flash-preview-image-generation'. LLM Reference model is lifecycle-deprecated; score kept as historical. Source: LMArena (CC BY 4.0). Retrieved 2026-10-17.
Attribution. Source: LMArena (lmarena.ai), CC BY 4.0
Scores rounded; model entries mapped to LLM Reference models; snapshot 2026-10-06.
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- The row names the exact model configuration, benchmark version, evaluator, harness, and observation date.
- Cost, token, and step averages are evaluated together with the score for the same configuration.
- You validate the configuration in your own Cursor workload before making a model choice.
Be cautious when
- Rows are collapsed into one score per base model or compared across benchmark versions.
- Small score differences are treated as decisive despite Cursor's variance caveat.
- A qualified row, including Grok 4.5's disclosed training-data advantage, is used as neutral winner evidence.
Related benchmarks
Last reviewed: Oct 17, 2026