Microsoft's MAI-Image-2.6 Lands at #2 on Text-to-Image Arena, Closing Gap on GPT Image 2
arena · x · 2026-08-11
Microsoft's MAI-Image-2.6 has officially launched on the Text-to-Image Arena leaderboard, securing the #2 spot overall with 1336 points.
The model trails the leading GPT Image 2 (Medium) by just 45 points and surpasses Grok Imagine Image 2.0 (Low) at #3 by 20 points. Showing massive gains over its predecessor MAI-Image-2.5 (ranked #10), the new version will be available on Playground next week, with early API access rolling out via Foundry.
Related event: Microsoft MAI-Image-2.6 Debuts at No. 2 on Text-to-Image Arena(4 posts)→
More from Multimodal
- Workflow: Generating Character Sheets Using MiniMax-H3 Video Model — JahJedi · 2026-08-11
- MiniMax H3 Generates Hilarious Pigeon Version of Tommy Wiseau — cocktailpeanut · 2026-08-11
- Meta Returns to Open Weights with Muse Glimmer (30B, Apache 2.0) — Simon Willison · 2026-08-11
- Developer Showcases Scene from Upcoming AI Video Generation Project — lmoroney · 2026-08-11
- Why Haven't Chinese Labs Distilled Suno and Released It as Open Weights? — breath_mirror · 2026-08-11
- Vizard Launches Video Agent: Turn Raw Footage or Ideas into Finished Videos Automatically — LearnWithBishal · 2026-08-11