openai / openai/codex

Sub-agent launcher rejects configured gpt-5.6-luna while direct Codex Luna run succeeds

Open
#38,344 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

app bug subagent
Dominant language
Rust
Stars
125k
Forks
19.5k
PR merge metrics
PR metrics pending

Description

Summary

The Codex multi-agent sub-agent launcher rejects gpt-5.6-luna as an unknown model even though Luna is configured for the Builder role and a direct Codex Luna Max run succeeds on the same machine and ChatGPT account.

Environment

  • macOS 26.5.1, arm64
  • Codex desktop-bundled CLI: 0.147.0-alpha.6.5
  • Authentication: ChatGPT
  • Global config: model = "gpt-5.6-luna", model_reasoning_effort = "max"
  • Global and project Builder definitions: model = "gpt-5.6-luna", model_reasoning_effort = "max"
  • Project sub-agent defaults also request Luna Max
  • Desktop app has been restarted several times without changing behavior

Reproduction

  1. From a GPT-5.6 task, create a sub-agent with the built-in Builder role, or explicitly request model gpt-5.6-luna with Max reasoning.
  2. The launcher rejects the request before the agent starts.

Observed errors:

Unknown model `gpt-5.6-luna` for spawn_agent. Available models: gpt-5.6-sol, gpt-5.6-terra

and the built-in Builder role cannot start because that role is fixed to Luna.

Control test

A direct ephemeral Codex run succeeds:

codex exec --ephemeral --json --skip-git-repo-check \
  -m gpt-5.6-luna \
  -c model_reasoning_effort='"max"' \
  -s read-only \
  "Reply with exactly LUNA_OK and nothing else."

Observed final agent message:

LUNA_OK

This suggests Luna is available to the account and local Codex runtime, but missing from the multi-agent launcher allowlist or capability snapshot for the active task.

Expected behavior

The sub-agent launcher should honor a valid configured Luna Max Builder, or return a specific explanation if Luna is intentionally unavailable for that sub-agent runtime.

Privacy

No repository code, credentials, account identifiers, or proprietary data are included in this report.

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start at the multi-agent sub-agent launcher and its spawn_agent model validation, then compare that path with the successful codex exec --ephemeral control command. Done means a configured gpt-5.6-luna Builder starts successfully, or the launcher provides a specific explanation when the model is unavailable.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
cli, devtools
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Quiet
Clarity
Mostly clear
Newbie friendliness
48/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.