[Feature]: local llms via codex (or claude)

[ogbinar]

Problem or motivation

Hello, i configured codex to use llama swap / llama cpp, and it is usable via codex but when i tried to change the model via telegram, it doesnt show the local models as an option. is this expected and if yes, can we request it as a feature? thank you

Proposed solution

expose non official models eg. local models configured via the ai tool

Alternatives considered

No response

Before submitting

• I have searched existing issues and feature requests for duplicates.

[ogbinar]

Or instead maybe we can connect to other cli tools like opencode

[PleasePrompto]

Status update: deferring to the next milestone. I spiked this during the Phase 3 provider work for v0.16.0 — the short answer is it's not the 1-line validator relaxation it looks like. What users actually want is a UI button for their local model in Telegram's /model selector, which means config schema + selector state + per-provider allow-lists + tests. It pairs well with #94 (Copilot) and the earlier PR #100 (OpenCode) in a dedicated "provider extensibility" phase. Keeping this open — will link to the tracking issue when that phase is scoped.