T3 Code vs Codex: model choice and customization

Jeff Escalante explains why T3 Code appeals to him as a coding assistant that resembles Codex but leaves more room for customization. He says he uses and likes Codex, and describes T3 Code as almost identical in the aspects he discusses. His preference centers on its open source code and model choice rather than a demonstrated difference in coding performance.

Escalante says T3 Code lets him select a cloud model in one thread and a GPT model in another. He adds that he can make this work in Codex, but doing so requires hacking it. The transcript does not identify the cloud model or show the configuration steps.

He also considers forking T3 Code to add his own status system. This is a possible modification, not a feature he says he has built. By contrast, he reports trying to patch the Codex app to suit his workflow and deciding the effort was not worthwhile. The comparison is a brief account of his preferences and customization attempts, without benchmarks or a setup walkthrough. It does not establish support for running models locally or offline.