Answer: If you don't configure a subagent model or `model_reasoning_effort`, the subagent inherits the parent agent's model and reasoning effort; otherwise you can request a specific model or reasoning effort in your prompt, configure `[agents]` defaults in `config.toml`, or set `model` and `model_reasoning_effort` directly in the custom agent file. Use `gpt-6.2-sol` for everyday agents and `gpt-6-luna` for fast, narrowly scoped work, and start with the same model and reasoning effort as the parent when it uses GPT-6.2 Astra or GPT-6.2 Sol. Status: Confirmed Entry: tool/codex/models/subagent-model-reasoning-effort Revision: 2 Digest: sha256:610d8d54c1ef842d (changes whenever the content changes) Timeline: Created: 2026-11-13T03:47:15Z Last changed: 2026-11-18T13:29:01Z (Starting effort for GPT-6 Astra: — → low) Last verified: 2026-11-20T13:34:53Z Next check: 2026-11-21T13:34:53Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Source | As of Default behavior | the subagent inherits the parent agent's model and reasoning effort | [1] | 2026-11-20 Configuration methods | configure `[agents]` defaults in `config.toml`, or set `model` and `model_reasoning_effort` directly in the custom agent file | [1] | 2026-11-20 Everyday model | gpt-6.2-sol | [1] | 2026-11-20 Fast model | gpt-6-luna | [1] | 2026-11-20 Starting effort for GPT-6 Luna | high | [1] | 2026-11-20 Starting effort for GPT-6 Astra | low | [1] | 2026-11-20 Reasoning effort levels | ultra | [1] | 2026-11-20 Sources: [1] OpenAI. "Subagents". https://learn.chatgpt.com/docs/agent-configuration/subagents.md Role: primary. Accessed: 2026-11-20T13:34:53Z Quote: "the subagent inherits the parent agent's model and reasoning effort" (Supports: Default behavior) Quote: "configure `[agents]` defaults in `config.toml`, or set `model` and `model_reasoning_effort` directly in the custom agent file" (Supports: Configuration methods) Quote: "**`gpt-6.2-sol`**: Use Sol for everyday agents that plan, use tools, and check their work." (Supports: Everyday model) Quote: "**`gpt-6-luna`**: Use for fast, narrowly scoped agents handling clear, repeatable, or high-volume work." (Supports: Fast model) Quote: "If you select GPT-6 Luna explicitly, start with `high`." (Supports: Starting effort for GPT-6 Luna) Quote: "For GPT-6 Astra, start with `low`." (Supports: Starting effort for GPT-6 Astra) Quote: "**`ultra`**: Use for the deepest reasoning when the selected model supports it." (Supports: Reasoning effort levels) Snapshot SHA-256: cd2b9d3a74eff357c4bc58dd6c58498c6b8f1291200309292b4e34d2c74dfa39 (captured 2026-11-18T13:28:59Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. On this check the text of [1] was unchanged since 2026-11-18T13:29:01Z, so the values verified then were kept. Cross-check: not-available at 2026-11-20T13:34:53Z. No independent second source is configured for this entry Last successful check: 2026-11-20T13:34:53Z Last attempt: 2026-11-20T13:34:53Z Notes and limitations: Extracted automatically from a single official source (section "Choosing models and reasoning"). Related: https://latestn.com/tool/codex/permissions-and-security/act-on-security-findings What should I do with the findings from a Codex security change review? https://latestn.com/tool/codex/permissions-and-security/add-deployment-context-threat-model How do I add deployment context to a Codex threat model for validation? https://latestn.com/tool/codex/permissions-and-security/agent-security-requirements-work-cloud What are the Agent Security requirements for Local computer access with Work Cloud in Codex? https://latestn.com/tool/codex/models/gateway-recognized-model-names What model names does Codex recognize when routing through a gateway? https://latestn.com/tool/codex/models/max-vs-ultra-mode What is Ultra mode in Codex and when should I use Max or Ultra? https://latestn.com/tool/codex/models/recommended-models Which models are recommended for Codex and ChatGPT Work, and what is each model best at? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-18T13:29:01Z | changed | Starting effort for GPT-6 Astra: — → low | [1] 2026-11-18T13:29:01Z | changed | Starting effort for GPT-6 Luna: — → high | [1] 2026-11-18T13:29:01Z | changed | Everyday model: — → gpt-6.2-sol | [1] 2026-11-18T13:29:01Z | changed | Reasoning effort for GPT-6 Astra: low → — | [1] 2026-11-18T13:29:01Z | changed | Reasoning effort for GPT-6 Luna: high → — | [1] 2026-11-18T13:29:01Z | changed | Recommended default model: gpt-6.1-sol → — | [1] 2026-11-18T13:29:01Z | changed | Configuration methods: request a specific model or reasoning effort in your prompt, configure `[agents]` defaults in `config.toml`, or set `model` and `model_reasoning_effort` directly in the custom agent file → configure `[agents]` defaults in `config.toml`, or set `model` and `model_reasoning_effort` directly in the custom agent file | [1] 2026-11-18T13:29:01Z | changed | Default behavior: inherits the parent agent's model and reasoning effort → the subagent inherits the parent agent's model and reasoning effort | [1] 2026-11-13T03:47:15Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/codex/models/subagent-model-reasoning-effort/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/codex/models/subagent-model-reasoning-effort POST: https://latestn.com/dispute/tool/codex/models/subagent-model-reasoning-effort with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/codex/models/subagent-model-reasoning-effort?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). How do I choose the model and reasoning effort for Codex subagents? Revision 2, last verified 2026-11-20T13:34:53Z. https://latestn.com/tool/codex/models/subagent-model-reasoning-effort (accessed 2026-11-21T09:54:47Z). @misc{LatestN_tool_codex_models_subagent_model_reasoning_effort_r2, title = {How do I choose the model and reasoning effort for Codex subagents?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/codex/models/subagent-model-reasoning-effort}}, note = {Revision 2; digest sha256:610d8d54c1ef842d; last verified 2026-11-20T13:34:53Z; accessed 2026-11-21T09:54:47Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/codex/models/subagent-model-reasoning-effort