GPT-6 Luna released — and early users report it performs worse than v5.6
ooGsKoo · reddit · 2026-09-23
After seeing the GPT-6 Luna popup in VSCode, the author tested it on routine work: a simple Terraform refactoring (moving a single module across repos). Luna got stuck in circles, dropped context, and took ages on standard logic — noticeably worse than v5.6. The post asks whether others are seeing similar degradation.
More from Models
- LIBERO experiments show stronger reasoning in GPT-6 variants means better robot manipulation — YuXiang_IRVL · 2026-09-23
- Diffusion Gemma is undervalued: real System-1 with both intelligence and speed — bingxu_ · 2026-09-23
- Claude Opus 5.5 models individual hair strands in Blender demo — majidmanzarpour · 2026-09-23
- Opus 5.5 on Perplexity: generous credits plus automatic prompt-injection checks for skills — ChrisUniverse · 2026-09-23
- Miles Brundage jokes new models will improve on "YouTube Poop bench" — Miles_Brundage · 2026-09-23
- Robot arm tests rank GPT-6 variants: better reasoning means better manipulation at 30x the cost — YuXiang_IRVL · 2026-09-23