OpenAI engineer: GPT-5 felt slow not due to latency but because it emitted too many tokens
npew · x · 2026-10-29
An OpenAI engineer recounts that after GPT-5's launch, user feedback consistently said "love the model, so slow." Investigation showed the bottleneck wasn't time-to-first-byte — the model simply generated too many tokens. The fix came from a full model generation spent obsessing over token efficiency, cutting output length to reduce perceived latency.
More from Models
- ARC nearing ceiling effect: Greg Kamradt proposes difficulty-ranked Elo list for public evals — GregKamradt · 2026-10-29
- Leaker: Google shipped Carbon ahead of OpenAI and Anthropic's next-gen models, still ~2 months behind — scaling01 · 2026-10-29
- ARC team: v1 hits ceiling effect as v2 extends headroom for frontier models — MindsAI_Jack · 2026-10-29
- Gemini 4 Argon Posts Lowest Hallucination Rate at 4% on AA-Omniscience — haider1 · 2026-10-29
- Suhail launches Particle: particle-1 matches performance with ~45% fewer tokens — chrisalbon · 2026-10-29
- Exploring a whole repo with a 1M-context open model cost just $0.63 — HowDevelop · 2026-10-29