FRAMEWIREIndonesiaUpdated Aug 29Live wire
0:00 / 0:00

Made this weapon configurator with Opus 5

Started with an image and asked Opus to recreate the gun using img2threejs, then turn it into an interactive customization experience

Dilum SanjayaAug 2976
0:00 / 0:00

Google's gemma 4 scored 5 out of 5 on every single prompt he threw at it running locally, coding, math

Reasoning, all of it, while opus 5 bills $200/MO and GPT-5.6 bills $200/MO for the same answers 01:28 he scrolls his own eval grid and reads it off, "five out of five, five

AntidAug 2910
0:00 / 0:00

This quietly ended the entire ai debate

Moon dev: "gpt 5.6 is the best at math and i can prove it in a way nobody expected" "i'm a die hard claude guy so watch me admit the thing i really did not want to admit" "opus 5 versus gpt 5.6, one versus two, and the gap is not what

Moon DevAug 29
0:00 / 0:00

Just imagine, $100K is on the line and Claude vs GPT have to decide where it goes.

Found Compelle instead of asking one AI a question and getting a confident answer, Compelle makes two AI agents take opposite sides and fight it out and the whole thing runs on Bittensor SN82

hammertimeAug 2928