[mrtesthah]

Gemini 3.5 is imminent.

https://www.voxelbench.ai/leaderboard

[londons_explore]

Getting a text based model to design voxel art really doesn't seem to be a good benchmark...

Better leaderboard: https://arena.ai/leaderboard/text

And the order of the top players is very different...

[mrtesthah]

All the frontier LLMs are multi-modal.

[londons_explore]

But voxel-art isn't one of their modes.

For other modes like image generation, the model has a dedicated image head, and is jointly trained.