Gemini Neon Claims 2-3x Rust Speedups Without Ever Benchmarking

bdsqlsz · x · 2026-11-22

bdsqlsz quotes @KIRASTUDI0's test showing Gemini app-tier models lag behind the Antigravity versions. The test: write a Rust algorithm and ask both AIs to make it 40% faster.

The poster argues this exposes Gemini models' core flaw: turning guesses into confident assertions, and that a model admitting uncertainty is far more useful than one that confidently bullshits you.

Original post →

More from Models

Models channel →