Skip to content

Unstable behavior of OpenAI models, via ChatGPT subscription, related to subagents #1151

Open

Description

@Ttimofeyka

Summary

When using OpenAI models (in particular, GPT-6.1 Sol or GPT-6.1 Terra) via ChatGPT Plus (via /connect), problems arise with the spawning of subagents, but only when another model is installed for subagents. In this case, the subagents are terminated immediately (0s | 0 tokens).

Expected Behavior

The usual stable spawn of subagents (as is the case with the usage of a single model).

Actual Behavior

The subagents appeared for a moment, and immediately were terminated.

Steps to reproduce the issue

  1. Set the planning + implementation model as GPT (through OpenAI subscription using /connect)
  2. For subagents, set the Deepseek V4.1 Flash 1017 as the model.
  3. Try to complete a usual task (possibly with telling the agent to spawn subagents for this task).

Command Code Version

1.89.1

Operating System

Linux

Terminal/IDE

Unknown

Shell

bash

Session file (optional)

No response

Fix prompt (optional)

No response

Additional context

No response

Activity

  1. added theissue type on Nov 16, 2026
  2. vipulgupta2048 commented on Nov 17, 2026

    @vipulgupta2048
    Member

    Thanks for the issue @Ttimofeyka Investigating this further.

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

    No labels
    No labels

    Type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions