11 replies

LMStudio (Local+Multiple models) + VM linux configuration help?

✅Solved

Hello, there!

I want to setup LMStudio bridge to local VM with linux to isolate the openclaw and create a local environment for him to isolate my personal data.
At the moment I'm fighting with Openclaw configuration file and I just can't handle it (I've tried ChatGPT and other AI to generate the config, but it just don't work).
My VM is configured flawlessly and I can ping my LMStudio from the VM - this shouldn't be the problem.
How do I setup openclaw for using multiple models?

At the moment I can't handle even 1 model locally

Here's the error and config I have in there:

  "models": {
    "providers": {
      "lmstudio": {
        "baseUrl": "http://XXX.XXX.122.1:1234/v1",
        "api": "openai-completions",
        "models": [
          {
            "id": "qwen/qwen3-coder-30b",
            "name": "Qwen3 Coder 30b",
            "api": "openai-completions",
            "contextWindow": 201033,
            "reasoning": false,
            "input": [
              "text"
            ],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "maxTokens": 8192
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "qwen/qwen3-coder-30b"
      },
      "models": {
        "qwen/qwen3-coder-30b": {}
      },
      "workspace": "/home/XXXX/.openclaw/workspace",
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },

  "models": {
    "providers": {
      "lmstudio": {
        "baseUrl": "http://XXX.XXX.122.1:1234/v1",
        "api": "openai-completions",
        "models": [
          {
            "id": "qwen/qwen3-coder-30b",
            "name": "Qwen3 Coder 30b",
            "api": "openai-completions",
            "contextWindow": 201033,
            "reasoning": false,
            "input": [
              "text"
            ],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "maxTokens": 8192
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "qwen/qwen3-coder-30b"
      },
      "models": {
        "qwen/qwen3-coder-30b": {}
      },
      "workspace": "/home/XXXX/.openclaw/workspace",
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },

Error code:

3:59:18 AM
error
diagnostic
lane task error: lane=main durationMs=13 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
diagnostic
lane task error: lane=session:agent:main:main durationMs=15 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
Embedded agent failed before reply: Unknown model: qwen-portal/qwen3-coder-30b

3:59:18 AM
error
diagnostic
lane task error: lane=main durationMs=13 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
diagnostic
lane task error: lane=session:agent:main:main durationMs=15 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
Embedded agent failed before reply: Unknown model: qwen-portal/qwen3-coder-30b

Solution

@Yury your config is almost fine — the error is coming from model naming, not networking.

Why you’re getting

Unknown model: qwen-portal/qwen3-coder-30b

Unknown model: qwen-portal/qwen3-coder-30b

OpenClaw parses model refs as:

- <provider>/<modelId>
<provider>/<modelId>
(it splits on the first

)

So when you set:

-

primary: "qwen/qwen3-coder-30b"

primary: "qwen/qwen3-coder-30b"

…OpenClaw thinks the provider is

qwen

qwen

(which is an alias for qwen-portal
qwen-portal
), and the modelId is

qwen3-coder-30b

qwen3-coder-30b

. Since you didn’t configure the Qwen provider, you get:

-

Unknown model: qwen-portal/qwen3-coder-30b

Unknown model: qwen-portal/qwen3-coder-30b

Fix: prefix it with your LM Studio provider (

lmstudio

lmstudio

)

Because your LM Studio model id itself contains a

(

qwen/qwen3-coder-30b

qwen/qwen3-coder-30b

), your OpenClaw ref must be:

Jump to solution

FotC�

Friends of the Crustacean 🦞🤝•6d ago•

11 replies

Yury

LMStudio (Local+Multiple models) + VM linux configuration help?

✅Solved

Here's the error and config I have in there:

  "models": {
    "providers": {
      "lmstudio": {
        "baseUrl": "http://XXX.XXX.122.1:1234/v1",
        "api": "openai-completions",
        "models": [
          {
            "id": "qwen/qwen3-coder-30b",
            "name": "Qwen3 Coder 30b",
            "api": "openai-completions",
            "contextWindow": 201033,
            "reasoning": false,
            "input": [
              "text"
            ],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "maxTokens": 8192
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "qwen/qwen3-coder-30b"
      },
      "models": {
        "qwen/qwen3-coder-30b": {}
      },
      "workspace": "/home/XXXX/.openclaw/workspace",
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },

  "models": {
    "providers": {
      "lmstudio": {
        "baseUrl": "http://XXX.XXX.122.1:1234/v1",
        "api": "openai-completions",
        "models": [
          {
            "id": "qwen/qwen3-coder-30b",
            "name": "Qwen3 Coder 30b",
            "api": "openai-completions",
            "contextWindow": 201033,
            "reasoning": false,
            "input": [
              "text"
            ],
            "cost": {
              "input": 0,
              "output": 0,
              "cacheRead": 0,
              "cacheWrite": 0
            },
            "maxTokens": 8192
          }
        ]
      }
    }
  },
  "agents": {
    "defaults": {
      "model": {
        "primary": "qwen/qwen3-coder-30b"
      },
      "models": {
        "qwen/qwen3-coder-30b": {}
      },
      "workspace": "/home/XXXX/.openclaw/workspace",
      "maxConcurrent": 4,
      "subagents": {
        "maxConcurrent": 8
      }
    }
  },

Error code:

3:59:18 AM
error
diagnostic
lane task error: lane=main durationMs=13 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
diagnostic
lane task error: lane=session:agent:main:main durationMs=15 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
Embedded agent failed before reply: Unknown model: qwen-portal/qwen3-coder-30b

3:59:18 AM
error
diagnostic
lane task error: lane=main durationMs=13 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
diagnostic
lane task error: lane=session:agent:main:main durationMs=15 error="Error: Unknown model: qwen-portal/qwen3-coder-30b"
3:59:18 AM
error
Embedded agent failed before reply: Unknown model: qwen-portal/qwen3-coder-30b

Solution

@Yury your config is almost fine — the error is coming from model naming, not networking.

Why you’re getting

Unknown model: qwen-portal/qwen3-coder-30b

Unknown model: qwen-portal/qwen3-coder-30b

OpenClaw parses model refs as:

- <provider>/<modelId>
<provider>/<modelId>
(it splits on the first

)

So when you set:

-

primary: "qwen/qwen3-coder-30b"

primary: "qwen/qwen3-coder-30b"

…OpenClaw thinks the provider is

qwen

qwen

(which is an alias for qwen-portal
qwen-portal
), and the modelId is

qwen3-coder-30b

qwen3-coder-30b

. Since you didn’t configure the Qwen provider, you get:

-

Unknown model: qwen-portal/qwen3-coder-30b

Unknown model: qwen-portal/qwen3-coder-30b

Fix: prefix it with your LM Studio provider (

lmstudio

lmstudio

)

Because your LM Studio model id itself contains a

(

qwen/qwen3-coder-30b

qwen/qwen3-coder-30b

), your OpenClaw ref must be:

Jump to solution

LMStudio (Local+Multiple models) + VM linux configuration help?

Why you’re getting
`Unknown model: qwen-portal/qwen3-coder-30b`
`Unknown model: qwen-portal/qwen3-coder-30b`

Fix: prefix it with your LM Studio provider (
`lmstudio`
`lmstudio`
)

LMStudio (Local+Multiple models) + VM linux configuration help?

Why you’re getting
`Unknown model: qwen-portal/qwen3-coder-30b`
`Unknown model: qwen-portal/qwen3-coder-30b`

Fix: prefix it with your LM Studio provider (
`lmstudio`
`lmstudio`
)

Similar Threads

Similar Threads

Similar Threads

LMStudio (Local+Multiple models) + VM linux configuration help?

Why you’re getting Unknown model: qwen-portal/qwen3-coder-30bUnknown model: qwen-portal/qwen3-coder-30b

Fix: prefix it with your LM Studio provider (lmstudiolmstudio)

LMStudio (Local+Multiple models) + VM linux configuration help?

Why you’re getting Unknown model: qwen-portal/qwen3-coder-30bUnknown model: qwen-portal/qwen3-coder-30b

Fix: prefix it with your LM Studio provider (lmstudiolmstudio)

Similar Threads

Similar Threads

Similar Threads

Why you’re getting
`Unknown model: qwen-portal/qwen3-coder-30b`
`Unknown model: qwen-portal/qwen3-coder-30b`

Fix: prefix it with your LM Studio provider (
`lmstudio`
`lmstudio`
)

Why you’re getting
`Unknown model: qwen-portal/qwen3-coder-30b`
`Unknown model: qwen-portal/qwen3-coder-30b`

Fix: prefix it with your LM Studio provider (
`lmstudio`
`lmstudio`
)