Three token-saving prompts, one each for Astra and the Sol models

rari @0xwhrrariOriginally posted 2026.10.08

Open original on X

These 3 prompts saved me $1000/month in wasted tokens One for GPT-6 Astra, one for GPT-6.1 Sol, one for GPT-6 Sol No full-repo reads for small tasks, no duplicate searches, no endless "just in case" checks Goal → Relevant context → Smallest useful action → Real check → Stop Pick your model, send it the matching prompt and make every token earn its place Then read my full guide to making GPT-6 Astra earn its price below ↓

This post quotes another post: “When to use GPT-6 Astra and when Sol is enough”