I couldn't find a good arena or leaderboard for image to 3D AIs... ...so I just made my own. Inspired by LMSYS, ImgSYS, and text-to-image arenas, the arena does head-to-head match-ups for image-to-3D genai. The leaderboard calculates the ELO score. Open to suggestions/submissions for both input images as well as image-to-3D models and products that should be included.

@Shpigford Just focus on api docs there are already good tools for repos and regular websites

@lmstudio I cannot decode this speculation

@NeginRaoof_ Looking forward to seeing swe-bench there! How about adding the aider polyglot benchmark?

@MatthewBerman yes please

@anton_lozhkov @Promptmethus i'd be curious to see things that involve spatial reasoning. this is a classic pitfall of llms (even vllms).

@anton_lozhkov cc @PrimeIntellect and @Muennighoff

@kepano AI e2e testing as the first pass review. This is easy now with packages like “browser-use” @browser_use

@jonasgeiping really cool stuff! i'm personally most excited about this part: "can capture types of reasoning that are not easily represented in words" i think this will be really powerful for physical sciences and engineering, especially for things that require spatial reasoning.

@jonas “My side-quest for the year is to spend a whole day coding while walking 20k steps on my treadmill desk without using my keyboard.” GOGOGO! What a noble quest. My version would be outside talking to my phone, perhaps via Remote Desktop. But I will need your magic spells first!

@iamrobertopablo @karpathy @levelsio @aquavoice_ I have a feeling the aqua team is like cursor - way too cool to document their product features

@ozgrozer @msfeldstein @cursor_ai @FAL +1 for both of y’all to publish your MCPs (fal & replicate) The bar is really low guys, more than half of MCP servers don’t work ;)

@karpathy @levelsio seriously try @aquavoice_ guys. superwhisper is cool but aquavoice => cursor composer is hands free vibe coding with the ability to dictate edits and aqua can learn the custom vocabulary directly from whatever app is in focus.

@levelsio @superwhisperapp i think @aquavoice_ is better, specifically because of the ability to: 1) detect special words from the app (if you allow that permission) and 2) you can just dictate a redo if it misunderstands or you say something wrong.

@tom_doerr Wow! There was already no moat in software. Now it’s really gone

I’m too lazy to read your docs, so how hard are you going to make it for my AI agent to learn how use your product/framework/package?

@msfeldstein @FAL ( )

@msfeldstein @FAL lol. not surprised. i'm so stoked for this! please do publish.

@msfeldstein i love the new ffmeg endpoint! so cool.

@kevinakwok this is genius