Skip to content

gpt-5.6-sol drops Codex Desktop thread-management tools that are available on gpt-5.6-terra in the same task #33637

Open

Description

@amatsuki5032

What version of Codex is running?

  • Codex Desktop for Windows: 26.707.12708.0
  • Bundled/local CLI observed: codex-cli 0.144.1
  • Subscription: ChatGPT Pro
  • OS: Windows x64
  • Host mode: local task, Full Access

What issue are you seeing?

In the same persistent Codex Desktop task, switching only the model from gpt-5.6-terra to gpt-5.6-sol removes the model-visible codex_app thread-management tool surface. Switching back to Terra restores it.

This affects system-owned Codex Desktop tools, not a custom MCP server.

Affected tools include:

  • codex_app.list_threads
  • codex_app.read_thread
  • codex_app.send_message_to_thread
  • codex_app.create_thread
  • codex_app.fork_thread
  • codex_app.set_thread_title
  • codex_app.set_thread_archived
  • codex_app.set_thread_pinned
  • codex_app.handoff_thread

On the Sol turn, only these codex_app tools were available:

  • navigate_to_codex_page
  • read_thread_terminal
  • load_workspace_dependencies

Controlled same-task reproduction

Technical task ID: 019f6bb0-7f30-7b61-938c-7b9843d6ec18

The task, cwd, permissions, and reasoning level stayed the same. Only the model changed.

  1. Start/use the task with gpt-5.6-terra, reasoning low.
  2. Ask which listed codex_app thread tools are available.
  3. Terra reports them available. A direct inspection of the current tool registry returned all nine listed tools.
  4. Switch the same task to gpt-5.6-sol, reasoning low.
  5. Ask again. Sol receives only the three basic codex_app tools listed above and cannot call list_threads, read_thread, send_message_to_thread, or create_thread.
  6. Switch the same task back to gpt-5.6-terra.
  7. Direct registry inspection again returns all nine tools.

Observed sequence:

gpt-5.6-terra -> 9 thread-management tools available
gpt-5.6-sol   -> thread-management tools absent; only 3 basic app tools
gpt-5.6-terra -> 9 thread-management tools available again

The local rollout records the model changes and results at approximately:

  • Terra turn: 2026-07-16T16:09:27Z
  • Sol turn: 2026-07-16T16:12:08Z
  • Terra control: 2026-07-16T16:21:54Z

Expected behavior

If these thread-management tools are enabled for the user/task, changing from Terra to Sol should not silently remove them. If Sol intentionally does not support them, the limitation should be documented and surfaced in the UI before model switching.

Actual behavior

The callable tool surface changes solely with the selected model. A long-lived orchestrator task can read/create/message other tasks on Terra, then loses those capabilities immediately on Sol, and regains them when switched back.

Impact

This breaks persistent multi-thread orchestration on the strongest reasoning model. Parent/router tasks cannot reliably:

  • receive or inspect worker results;
  • send completion callbacks or follow-up instructions;
  • create worker tasks;
  • archive, pin, or rename managed tasks.

It also makes the failure look intermittent or focus/thread dependent when it is reproducibly model-turn dependent in this test.

Suspected boundary

The tool definitions exist in the Desktop installation and are admitted for Terra, but are not selected/registered into the Sol turn's model-visible tool surface. This looks like a model-route/tool-registry construction issue, potentially related to the Responses Lite/deferred-tool exposure class, rather than a missing local configuration.

Related:

I can provide the local rollout excerpt or screenshots if maintainers need them. No repository data is required to reproduce the model A/B/A switch.

Activity

  1. amatsuki5032 commented on Jul 16, 2026

    @amatsuki5032
    Author

    Additional visual and same-task confirmation:

    After the Sol turn exposed only:

    • navigate_to_codex_page
    • read_thread_terminal
    • load_workspace_dependencies

    the user switched the same task from GPT-5.6 Sol back to GPT-5.6 Terra. The very next Terra turn reported and directly observed all nine thread-management tools again.

    The Desktop UI visibly records the model transition (GPT-5.6 Sol -> GPT-5.6 Terra) between the two responses. No restart, new task, permission change, cwd change, or task opening/focus workaround was needed.

    This strengthens the boundary to a reversible per-turn model-route difference:

    Terra: thread tools present
    Sol:   thread tools absent
    Terra: thread tools present again
    

    One Terra response described the preceding Sol result as a mistaken inspection, but the persisted rollout and the actual tool registries disagree with that explanation: the Sol turn really received only the three basic Codex app tools, while both Terra turns received the full thread-management set.

  2. added
    bugSomething isn't working
    appIssues related to the Codex desktop app
    app-serverIssues involving app server protocol or interfaces
    on Jul 16, 2026
  3. github-actions commented on Jul 16, 2026

    @github-actions
    Contributor
  4. KanadeK commented on Sep 26, 2026

    @KanadeK

    Additional Windows/Desktop reproduction on 2026-09-26, related to this model-dependent tool exposure issue and #46147.

    Environment: Codex Desktop 26.924.1866.0 (x64), codex-cli 0.158.0-alpha.2, Windows 11 build 26200. The user has fully restarted the Desktop app; the problem persisted.

    In the affected GPT-6 conversation, the app guidance still directs the agent to use the native Codex chat/thread tools, but the live callable catalog contains none of create_thread, list_threads, read_thread, send_message_to_thread, fork_thread, or wait_threads. There is also no tool_search route that can expose them. This is an absent-tool condition, not an invocation returning permission denied. The agent can use generic subagent tools, but those cannot create or message user-owned Desktop chats.

    The user reports that a GPT-5.6 Sol conversation in the same Desktop installation/account has been able to perform cross-chat creation and messaging. I did not perform a controlled same-thread model A/B/A switch in this diagnostic, so please treat that comparison as user-observed rather than a proven causal model switch. The current GPT-6 tool-catalog check was performed directly.

    Repro/impact: ask an affected GPT-6 Desktop chat to inspect an existing sidebar chat, create a new visible chat, or send a message to another existing chat. The requested native tool is absent, preventing coordination of persistent, user-owned work. Restarting the app did not restore the capability.

    Expected: expose the native tools consistently when the account/session permits them, or return a clear structured reason when unavailable. It would help to inspect model-route tool registration and the effective codex_app/plugin tool allowlist for GPT-6 versus GPT-5.6 Sol.

    No private chat IDs, repository names, local paths, credentials, or raw configuration/log content are included.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    appIssues related to the Codex desktop appapp-serverIssues involving app server protocol or interfacesbugSomething isn't working

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions