Image Editing Arena Revamp: MAI-Image-2.6 Leads, GPT Image 2 Best at Local Edits
ArtificialAnlys · x · 2026-09-03
Artificial Analysis has overhauled its Image Editing Arena to test complex, chained edit instructions across 7 editing actions and 10 real-world use cases, from Marketing & Advertising to UI/UX Design.
Key findings:
- MAI-Image-2.6-Preview tops the overall leaderboard and leads 4 of 7 editing actions (Scene & Style, Text/Symbol Edits, Reasoning-Based Edit, Enhancement & Restoration), excelling at restyling, relighting, and retouching.
- GPT Image 2 (high) ranks #2 overall but #1 on Object-Level Edit and Composition & Framing — strongest for precise local edits and reframing; weaker at whole-image transformations.
- Seedream 5.0 Pro is the identity-preservation specialist, #1 at keeping character likeness, outfit, and accessories consistent ($90 per 1,000 images).
- MAI-Image-2.5-Flash is the value pick at $20 per 1,000 images vs. $211 for GPT Image 2 (high).
The analysis argues editing is no longer a side feature: frontier models have saturated single-instruction edits, and multi-step chained instructions plus per-use-case fit are the new differentiators. An upcoming Reference-to-Image benchmark will cover generation from reference images as a separate workflow stage.
More from Multimodal
- Luma shares filmmaking-style workflow: motion-capture real stunts then restyle with video-to-video — lmoroney · 2026-10-18
- MOSS-Image-SFT at 4 Steps Draws Hype: 320x320 to 2K Without Artifacts — Powerful_Evening5495 · 2026-10-18
- MOSS SFT vs Flux 3 Klein vs SD 3.0: Same-Prompt Comparison Exposes SD3's Age — ZootAllures9111 · 2026-10-18
- Dev builds Mumbai-inspired 3D manga background generator from OpenStreetMap data — CurieuxExplorer · 2026-10-18
- Omni 1'1 flash video demo generates atmospheric foggy Victorian haunted house — Dense-Instruction210 · 2026-10-18
- Flux 3 klein distilled tested: int8 workflow delivers surprisingly good 2MP images — Ok_Tale7582 · 2026-10-18