[Bug]: GPT-5.6 Luna works in Codex CLI after quota exhaustion, but fails in T3 Code

Before submitting

• I searched existing issues and did not find a duplicate.
• I included enough detail to reproduce or investigate the problem.

Area

apps/server

Steps to reproduce

Exhaust the normal Codex usage quota while Luna Reserve is still available.
Open the official Codex CLI.
Select:
Model: gpt-5.6-luna
Reasoning: medium
Send a prompt.
Observe that the request succeeds.
Open T3 Code using the same Codex account.
Select:
Model: gpt-5.6-luna
Reasoning: medium
Send a prompt.

Expected behavior

T3 Code should behave consistently with the official Codex CLI and use the available Luna Reserve capacity when normal Codex quota is exhausted.

Actual behavior

The official Codex CLI successfully responds using gpt-5.6-luna with Medium reasoning, while T3 Code fails because the regular Codex quota is exhausted.

Impact

Blocks work completely

Version or commit

No response

Environment

Provider: Codex
Model: gpt-5.6-luna
Reasoning: medium
Same machine/account tested in Codex CLI and T3 Code
Normal Codex quota exhausted
Luna Reserve still available

Logs or stack traces

Screenshots, recordings, or supporting files

No response

Workaround

No response

Start in apps/server by tracing how Codex requests select models and handle exhausted normal quota, then compare that behavior with the official Codex CLI for gpt-5.6-luna at medium reasoning. Reproduce with normal Codex quota exhausted and Luna Reserve available; done means T3 Code succeeds under the same conditions.