[Bug]: Subtasks with gpt-5.4 crashes at start or mid-run #2428
Description
Prerequisites
- I will write this issue in English (see our Language Policy)
- I have searched existing issues to avoid duplicates
- I am using the latest version of oh-my-opencode
- I have read the documentation or asked an AI coding agent with this project's GitHub URL loaded and couldn't find the answer
Bug Description
So, i started using the new 3.11 version with gpt 5.4 (trought azure openai), and it really breaks the flow, interrupts before doing anything(look image):
and it also seems the atlas thinks the task is still going on.
other times it happened the agent would get completely stopped mid-run and/or mid final response, and it would correctly be saw as completed, even if it wasnt actually
This seems to happen ONLY with gpt-5.4
Steps to Reproduce
configure opencode with gpt-5.4
run whatever agent(expecially subagents)
you probably will not be able to execute a plan or plan one with prometeus without having problems
Expected Behavior
Just work like any other agent without those bugs
Actual Behavior
.
Doctor Output
{
"plugin": ["oh-my-opencode@latest"],
"$schema": "https://opencode.ai/config.json",
"provider": {
"azure-anthropic": {
"name": "Azure AI Foundry (Anthropic)",
"npm": "@ai-sdk/anthropic",
"api": "https://[REDACTED_RESOURCE_NAME].services.ai.azure.com/anthropic/v1",
"env": ["AZURE_ANTHROPIC_API_KEY"],
"models": {
"claude-opus-4-6": {
"id": "claude-opus-4-6",
"name": "Claude Opus 4.6",
"tool_call": true,
"attachment": true,
"reasoning": true,
"temperature": true
},
"claude-sonnet-4-5": {
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5",
"tool_call": true,
"attachment": true,
"reasoning": true,
"temperature": true
},
"claude-sonnet-4-6": {
"id": "claude-sonnet-4-6",
"name": "Claude Sonnet 4.6",
"tool_call": true,
"attachment": true,
"reasoning": true,
"temperature": true
},
"claude-haiku-4-5": {
"id": "claude-haiku-4-5",
"name": "Claude Haiku 4.5",
"tool_call": true,
"attachment": true,
"reasoning": true,
"temperature": true
}
}
},
"azure": {
"models": {
"gpt-5.4": {
"id": "gpt-5.4",
"name": "GPT-5.4",
"npm": "@ai-sdk/openai",
"api": "https://${AZURE_RESOURCE_NAME}.cognitiveservices.azure.com/openai/v1",
"env": ["AZURE_OPENAI_API_KEY"]
}
}
},
"google": {
"models": {
"gemini-3.1-pro-preview": {
"id": "gemini-3.1-pro-preview",
"name": "Gemini 3 Pro Preview",
"npm": "@ai-sdk/google",
"env": ["GEMINI_API_KEY"]
},
"gemini-3-flash-preview": {
"id": "gemini-3-flash-preview",
"name": "Gemini 3 Flash Preview",
"npm": "@ai-sdk/google",
"env": ["GEMINI_API_KEY"]
}
}
}
},
"mcp": {
"context7": {
"type": "remote",
"url": "https://mcp.context7.com/mcp",
"headers": {
"CONTEXT7_API_KEY": "[REDACTED_API_KEY]"
},
"enabled": true
},
"gitnexus": {
"type": "local",
"command": ["npx", "-y", "gitnexus@latest", "mcp"],
"enabled": true
}
}
}Error Logs
Configuration
{
"$schema": "https://raw.githubusercontent.com/code-yeongyu/oh-my-opencode/master/assets/oh-my-opencode.schema.json",
"google_auth": false,
"disabled_mcps": ["chrome-devtools"],
"agents": {
"sisyphus": {
"model": "azure-anthropic/claude-opus-4-6",
"variant": "high"
},
"oracle": {
"model": "azure/gpt-5.4",
"variant": "high"
},
"librarian": {
"model": "google/gemini-3-flash-preview",
"variant": "high"
},
"explore": {
"model": "google/gemini-3-flash-preview",
"variant": "high"
},
"multimodal-looker": {
"model": "azure/gpt-5.3-codex",
"variant": "high"
},
"prometheus": {
"model": "azure-anthropic/claude-opus-4-6",
"variant": "max"
},
"metis": {
"model": "azure-anthropic/claude-opus-4-6",
"variant": "high"
},
"momus": {
"model": "azure/gpt-5.4",
"variant": "high"
},
"atlas": {
"model": "azure-anthropic/claude-opus-4-6",
"variant": "high"
},
"build": {
"model": "azure-anthropic/claude-opus-4-6",
"variant": "medium"
},
"sisyphus-junior": {
"model": "azure-anthropic/claude-sonnet-4-6",
"variant": "high"
},
"hephaestus": {
"model": "azure/gpt-5.3-codex",
"variant": "medium"
}
},
"categories": {
"quick": {
"model": "google/gemini-3-flash-preview",
"variant": "high",
"description": "Trivial tasks - single file changes, typo fixes, simple modifications"
},
"visual-engineering": {
"model": "google/gemini-3.1-pro-preview",
"variant": "high",
"description": "Frontend, UI/UX, design, styling, animation"
},
"ultrabrain": {
"model": "azure/gpt-5.3-codex",
"variant": "xhigh",
"description": "Deep logical reasoning, complex architecture decisions"
},
"deep": {
"model": "azure/gpt-5.3-codex",
"variant": "medium",
"description": "Autonomous problem-solving with thorough research"
},
"artistry": {
"model": "google/gemini-3.1-pro-preview",
"variant": "high",
"description": "Highly creative/artistic tasks, novel ideas"
},
"unspecified-low": {
"model": "azure/gpt-5.4",
"variant": "medium",
"description": "Tasks that don't fit other categories, low effort required" },
"unspecified-high": {
"model": "azure/gpt-5.4",
"variant": "high",
"description": "Tasks that don't fit other categories, high effort required" },
"writing": {
"model": "google/gemini-3-flash-preview",
"variant": "high",
"description": "Documentation, prose, technical writing"
}
}
}Additional Context
No response
Operating System
Linux
OpenCode Version
1.2.24