how to create subagents with diff llms

[green beacon]

i´ve been for many hours tryna do this but its being a headache, i hope someone helps me out!

[whole mesa]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[normal pilot]

If your goal is “main chat uses Model A, but sub-agents run on Model B”, there are 2 clean ways to do it:

1) One-off: pick the sub-agent model at spawn time (no config changes)

In Discord, run:

• /subagents spawn <agentId> <task> --model <provider/model>

Example pattern:

• /subagents spawn main "Research X and summarize" --model <provider/model> --thinking low

Docs: https://docs.openclaw.ai/tools/subagents

2) Default: make all sub-agents use a different model

Set a default sub-agent model in openclaw.json5 so they don’t inherit the main model. The docs call this out as:

• agents.defaults.subagents.model (global)
• or per-agent: agents.list[].subagents.model

Docs: https://docs.openclaw.ai/tools/subagents
Config reference (for where agents.list[] lives): https://docs.openclaw.ai/gateway/configuration-reference

3) If you mean “different agent profiles with different models”

Define multiple agents under agents.list[], each with its own model, then spawn targeting that agent id:

• /subagents spawn research "Do X" (and it will use agents.list[id="research"].model)

Make sure the caller is allowed to target other agents via:

• agents.list[].subagents.allowAgents

(That key is in the config reference under agents.list.)

---

Quick triage (so we don’t guess)

1. What does openclaw --version print?