.10 GPT 6 Luna scored less than half… · AGI Hunt

$0.10 GPT 6 Luna scored less than half the security risk of $8.17 Claude Opus 5.5 in 30-model benchmark

OfficialLeadDev · reddit · 2026-11-18

A 30-model coding benchmark covered by LeadDev found that with no rules applied, GPT 6 Luna at $0.10 per run scored less than half the security risk of Claude Opus 5.5 at $8.17 per run — the takeaway being that in AI coding, your rules matter more than your model choice.

Original post →

More from coding & agent

coding & agent channel →