Vaibhav warns against casually switching Codex's default models
In response to suggestions to switch the Codex Dreaming V3 memory system's default models to gpt-6-luna and gpt-6.1-terra, Vaibhav Srivastav warns: unless you truly know what you're doing, don't switch the models, because the prompts for these models are carefully calibrated for the default models.
Topics
View original postCodex's Dreaming V3 memory system uses GPT-5.6 Luna and Terra by default. Upgrade by copying the following into config.toml: extract_model = "gpt-6-luna" consolidation_model = "gpt-6.1-terra" Both models are more intelligent AND cheaper than their predecessors. It's better than a free lunch. It's like getting paid to eat lunch.
View original
Codex's Dreaming V3 memory system uses GPT-5.6 Luna and Terra by default. Upgrade by copying the following into config.toml: extract_model = "gpt-6-luna" consolidation_model = "gpt-6.1-terra" Both models are more intelligent AND cheaper than their predecessors. It's better than a free lunch. It's like getting paid to eat lunch.
Body
Big FYI: DO NOT change models unless you really know what you’re doing, the prompts for these models are carefully calibrated for the defaults.