mostly GLM-5-turbo for coding,

and gemini-3.6-flash for brainstorming problems

( why ? token cost and availability at work )