User Report: GPT-6.2 Models Demand Extremely Precise Prompts or Get Lazy
VraserX · x · 2026-11-18
A user reports on GPT-6.2 Astra and Sol: both are extremely sensitive to prompting — imprecise instructions make them lazy or off-target. The tip: refine your prompts with ChatGPT first, and the results become genuinely outstanding.
More from Models
- Codex refuses to paste slides text, sparking debate that alignment needs external control — gerardsans · 2026-11-18
- GPT 6.2 models called 'lazy': fisheye-filtered images passed off as built 3D scenes — lyraxana · 2026-11-18
- DeepMind's fhuszar: Claude cybersecurity guardrail is the new NaN, kills an afternoon — fhuszar · 2026-11-18
- User impressed by Claude Computer Use speed, switching from ChatGPT for that purpose — koltregaskes · 2026-11-18
- Google would rather speedrun the periodic table than name models normally — Hesamation · 2026-11-18
- $20 Claude Opus 5.5 vs free Hy4 Flash: one-shot dance test showdown — CodeByPoonam · 2026-11-18