Configuring OpenClaw Model Routing | Generated by AI

Home 2026.04 English 中文

---

Question: How to set the default model to openrouter/minimax/minimax-m2.7 and switch to local DeepSeek R1 in OpenClaw?

Answer:

You’re configuring OpenClaw model routing. There are two parts:

• Set default model → openrouter/minimax/minimax-m2.7
• Switch manually to local model → /model llama-server/DeepSeek-R1-0528-Qwen3-8B-Q4_K_M.gguf

Here are the correct ways to do it.

---

Method 1 — Set default model (recommended)

Edit your OpenClaw config file:

~/.openclaw/config.yaml

or project-level:

.openclaw/config.yaml

Set:

model: openrouter/minimax/minimax-m2.7

Now OpenClaw will default to MiniMax M2.7.

---

Method 2 — Switch model interactively

Inside OpenClaw chat:

/model llama-server/DeepSeek-R1-0528-Qwen3-8B-Q4_K_M.gguf

This switches to local DeepSeek R1 immediately.

To switch back:

/model openrouter/minimax/minimax-m2.7

---

Method 3 — CLI launch override

You can also start OpenClaw with:

openclaw --model openrouter/minimax/minimax-m2.7

or local:

openclaw --model llama-server/DeepSeek-R1-0528-Qwen3-8B-Q4_K_M.gguf

---

Method 4 — Workspace-level (per project)

Create:

.project-root/.openclaw/config.yaml

model: openrouter/minimax/minimax-m2.7

This is useful if:

• cloud model for coding
• local DeepSeek for offline
• different repos → different models

---

Verify current model

Run:

/model

OpenClaw prints current active model.

---

Typical best practice

Default:

openrouter/minimax/minimax-m2.7

Switch when needed:

/model llama-server/DeepSeek-R1-0528-Qwen3-8B-Q4_K_M.gguf

This gives:

• fast cloud reasoning
• cheap local fallback
• offline capability

Exactly what your snippet describes.

---

Back Donate