2137At other intelligence levels, ask for subagents explicitly when you want work2137At other intelligence levels, ask for subagents explicitly when you want work
2138delegated in parallel.2138delegated in parallel.
2139 2139
2140If you don't pin a model or `model_reasoning_effort`, Codex can choose a setup2140If you don't configure a subagent model or `model_reasoning_effort`, the
2141that balances intelligence, speed, and price for the task. It may favor `gpt-5.6-terra` for fast scans or a higher-effort `gpt-5.6` configuration for more demanding reasoning. When you want finer control, steer that choice in your prompt or set `model` and `model_reasoning_effort` directly in the agent file.2141subagent inherits the parent agent's model and reasoning effort. If an explicit
2142spawn request or an `[agents]` default selects a model without an
2143explicit or configured reasoning effort, the subagent uses that model's default
2144reasoning effort. To balance intelligence, speed, and price for each task,
2145request a specific model or reasoning effort in your prompt,
2146configure `[agents]` defaults in `config.toml`, or set `model` and
2147`model_reasoning_effort` directly in the custom agent file.
2148For example, use `gpt-5.6-terra` for fast scans or a higher-effort `gpt-5.6` configuration for more demanding reasoning.
2142 2149
2143For most tasks in Codex, start with2150For most tasks in Codex, start with
2144`gpt-5.6`. Use2151`gpt-5.6`. Use
2265- `developer_instructions`2272- `developer_instructions`
2266 2273
2267If a custom agent file sets `model` or `model_reasoning_effort`, the value in2274If a custom agent file sets `model` or `model_reasoning_effort`, the value in
2268the file takes precedence. Otherwise, Codex resolves each setting independently:2275the file takes precedence. Before applying the file, Codex resolves each setting
2269an explicit spawn value, then the corresponding `[agents]` default, then the2276from an explicit spawn value, then the corresponding `[agents]` default, then
2270parent's value. If a spawn selects a different model and neither an explicit nor2277the parent's value. If an explicit spawn request or an `[agents]` default
2271configured effort is present, Codex uses that model's default effort. Other2278selects a model and neither supplies a reasoning effort, Codex uses
2272session settings, such as `sandbox_mode`, `mcp_servers`, and `skills.config`,2279that model's default effort. A custom agent file that sets only `model`
2273inherit from the parent when the custom agent file omits them.2280preserves this previously resolved effort. Set `model_reasoning_effort` in the
2281file too if the selected model doesn't support that effort or you want a
2282different one. Other session settings, such as `sandbox_mode`, `mcp_servers`,
2283and `skills.config`, inherit from the parent when the custom agent file omits
2284them.
2274 2285
2275#### Global settings2286#### Global settings
2276 2287
17931 17942
17932Source: [Worktrees](https://learn.chatgpt.com/docs/environments/git-worktrees.md)17943Source: [Worktrees](https://learn.chatgpt.com/docs/environments/git-worktrees.md)
17933 17944
17934In the ChatGPT desktop app, worktrees let Codex run multiple independent chats in the same project without interfering with each other. For Git repositories, [scheduled tasks](https://learn.chatgpt.com/docs/automations) can run on dedicated background worktrees so they don't conflict with your ongoing work. In non-version-controlled projects, scheduled tasks run directly in the project directory. You can also start chats in a worktree manually and use Handoff to move a chat between Local and Worktree.17945Worktrees let Codex run multiple independent chats in the same project without interfering with each other. The repository, worktree, and commands remain on the computer or remote development environment that contains the project. You can work directly in the ChatGPT desktop app, or use [Remote](https://learn.chatgpt.com/docs/remote) in the ChatGPT mobile app to start, guide, approve, and review worktree chats on a connected computer.
17946
17947For Git repositories, [scheduled tasks](https://learn.chatgpt.com/docs/automations) can run on dedicated background worktrees so they don't conflict with your ongoing work. In non-version-controlled projects, scheduled tasks run directly in the project directory. You can also start chats in a worktree manually and use Handoff to move a chat between Local and Worktree.
17935 17948
17936Worktrees are available only in Codex in the ChatGPT desktop app. Select17949Worktrees don't run locally on your phone. With Remote, the mobile app
17937**Codex** before you start a chat in a worktree.17950controls Codex on your connected computer, where the repository and worktree
17951remain, or in the remote development environment that computer uses. The
17952desktop-specific instructions below apply on the connected computer.
17938 17953
17939#### What's a worktree17954#### What's a worktree
17940 17955
