WHAT NEW IN X11.06 FRI

AI collections from X · 23 collections · 69,480 posts

gemini-4IMAGE

Test: Gemini 4 Argon makes many mistakes on a compiler, Meta Muse cleaner

A user reports Gemini 4 Argon made limited progress and many mistakes building on a compiler, needing Fable to clean up, while Meta Muse Wonder Full made none in a similar test.

Eric Hartford
@QuixiAIArchived 17 days agoOriginally posted 2026.10.18
Test: Gemini 4 Argon makes many mistakes on a compiler, Meta Muse cleaner
Eric Hartford @QuixiAIOriginally posted 2026.10.18
Open original on X

I let Gemini 4 Argon try building on my compiler. He made some progress but made lots of mistakes. Fable is cleaning up after him now. Muse Wonder Full, under a similar test, made no mistakes. For developing the With compiler, Meta Muse Wonder Full is better than Gemini 4 Argon.

x.com/quixiai/status/2111773335629758897Archived 2026.10.19 10:00