gpt-6POST
Tiering Astra usage: high for planning, medium for orchestration, cheap models for execution
The post suggests using Astra high for planning and Astra medium for orchestration while handing execution to cheap fast models like glm 5.3 flash and gemini 3.8 flash, to stretch Astra limits on a Plus plan.
Haider.
@haider1Archived 2026.09.16Originally posted 2026.09.15
Haider. @haider1Originally posted 2026.09.15
GPT-6 Astra (high) for planning. Astra (medium) for orchestration. fast/cheap models like glm 5.3 flash and gemini 3.8 flash for execution. probably the best way to stretch your Astra limits without wasting high effort on everything if you're on a plus plan trust me
x.com/haider1/status/2099719860175716527Archived 2026.09.16 10:21