10 Rounds of "I Believe in You" Made GPT-6 Code Faster and Leaner
mohamedmansour · x · 2026-09-09
Developer mohamedmansour found GPT-6 Astra leaves a lot of performance on the table in its first pass, even when explicitly asked to optimize. After 10+ rounds of "You can do it, I believe in you," the code kept getting faster, leaner, and more memory-efficient — eventually matching what he'd write himself. He half-jokes about running a 100-iteration Ralph loop of "make it perform better," possibly spending $1,000 in credits to discover the final optimization is a full rewrite in assembly.
More from coding & agent
- KVMem pages agent context overflow to KV state, beating compaction on DeepSWE — omarsar0 · 2026-09-09
- Coding agents are the first product to compete with themselves — samgoodwin89 · 2026-09-09
- Local Qwen 27B on an RTX 3090 Builds a Playable Mario Clone From One Prompt — ChopSticksPlease · 2026-09-09
- Vibe coder's loop: make the model draw architecture diagrams, write specs, and iterate to production quality — Daniel_Farinax · 2026-09-09
- RSM agent memory hits 83% of full-context quality at 32% token cost on a 4k budget — dair_ai · 2026-09-09
- Tencent Hunyuan details Gander, an end-to-end full-duplex omni interaction agent — Tencent-Hunyuan · 2026-09-09