Large Language Models are AI program's that ingest massive amounts of text. By analyzing this data, they are able to figure out patterns and structures - they can translate texts; write poems, emails, newsletters, etc; answer questions, even strange and open question.

3D generative models do a similar thing but with geometry and textures instead of words - they take in text or images and spit out 3D objects; meshes, textures, the whole thing. Some are fast, some are detailed, some are both.

TuringScore.com keeps track of:

• Rank: The model's position on the leaderboard.
• Model: The name of the model.
• Elo/Score: For text models, a relative "score" with confidence interval - the higher the elo the better the model. For 3D models, an overall quality score.
• Licensing/Organization: For text models, whether the weight's are available and if they can be used commercially. For 3D models, who made it.

Data from LMSYS's Chatbot Arena and 3DTopia's 3DGen Leaderboard.