set-model

Set this organization's default AI model, the one every agent uses unless it names its own, or clear it. It affects every project in the organization, so it needs an admin role and the fabric:model.write grant.

curl -sS -X POST https://fabric.wexa.ai/mcp/$FABRIC_PROJECT_ID \
  -H "authorization: Bearer $FABRIC_API_KEY" \
  -H "Mcp-Session-Id: $MCP_SESSION_ID" \
  -H "content-type: application/json" -H "accept: application/json" \
  -d '{"jsonrpc":"2.0","id":3,"method":"tools/call","params":{"name":"set-model","arguments":{"clear":true}}}'

{
  "content": [
    {
      "type": "text",
      "text": "{\"lifecycle_id\":\"qlc_000172\",\"result\":{\"changed\":true,\"default\":\"\",\"previous\":\"mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8\"}}"
    }
  ],
  "isError": false
}

curl -sS https://fabric.wexa.ai/v1/set-model \
  -H "authorization: Bearer $FABRIC_API_KEY" \
  -H "content-type: application/json" \
  -d '{"clear":true}'

{
  "lifecycle_id": "qlc_000172",
  "result": {
    "changed": true,
    "default": "",
    "previous": "mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8"
  }
}

import { Fabric } from '@wexa-fabric/sdk'

const fabric = new Fabric()
const result = await fabric.setModel({
  clear: true
})

// result is the payload itself, with no `.result` to read
{
  changed: true,
  default: '',
  previous: 'mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8'
}

from wexa import Fabric

fabric = Fabric()
out = fabric.set_model(clear=True)

# out is the result, already unwrapped
{
    "changed": True,
    "default": "",
    "previous": "mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8"
}

Parameters

Parameter	Type	Required	Notes
`model`	string	Optional	Registry model id to make the organization default (`list-models` `settable`). Omit when clearing. One of `model` or `clear` is needed, not both. `system_model` and `default` are refused, as is an id the organization has not configured.
`clear`	boolean	Optional	Return the organization to the platform default instead of choosing a model.
`fallback`	string	Optional	A second model id used when the default is unavailable. Stored with the organization's other model settings, not in the registry. Refused when the organization has not configured it.

Errors

ValidationError (422): the credential does not hold an admin role (OWNER, ORG_ADMIN or PROJECT_ADMIN).

Show full response

A real response, captured from a live call to set-model:

{
  "content": [
    {
      "type": "text",
      "text": "{\"lifecycle_id\":\"qlc_000172\",\"result\":{\"changed\":true,\"default\":\"\",\"note\":\"the organization is back on the platform default. written to the AI Models registry, the same store the console's AI Models page shows and every execution resolves through. Agents that name their own `llm.model` are unaffected.\",\"previous\":\"mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8\",\"scope\":\"organization 6a79a770b6c128ccc24bfca8 \\u2014 every project in it\"}}"
    }
  ],
  "isError": false
}

{
  "lifecycle_id": "qlc_000172",
  "result": {
    "changed": true,
    "default": "",
    "note": "the organization is back on the platform default. written to the AI Models registry, the same store the console's AI Models page shows and every execution resolves through. Agents that name their own `llm.model` are unaffected.",
    "previous": "mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8",
    "scope": "organization 6a79a770b6c128ccc24bfca8 \u2014 every project in it"
  }
}

// the object you get back IS the result — there is no `.result` to read
{
  changed: true,
  default: '',
  note: 'the organization is back on the platform default. written to the AI Models registry, the same store the console\'s AI Models page shows and every execution resolves through. Agents that name their own `llm.model` are unaffected.',
  previous: 'mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8',
  scope: 'organization 6a79a770b6c128ccc24bfca8 — every project in it'
}

// The TS SDK names these camelCase — lifecycleId, traceId — where Python uses
// lifecycle_id and trace_id. Both are non-enumerable: readable, but omitted by
// Object.keys() and JSON.stringify(). `result.lifecycleId` here is undefined.
result.lifecycleId // 'qlc_000172'

# a dict subclass; `result` is already unwrapped
{
 "changed": True,
 "default": "",
 "note": "the organization is back on the platform default. written to the AI Models registry, the same store the console's AI Models page shows and every execution resolves through. Agents that name their own `llm.model` are unaffected.",
 "previous": "mdl_bx_anthropic-claude-haiku-4-5-20251001-v1_28ccc24bfca8",
 "scope": "organization 6a79a770b6c128ccc24bfca8 \u2014 every project in it"
}

out.lifecycle_id  # 'qlc_000172' — an attribute, not a key

More details

In the console, open Settings, then Org, then AI Models. Open the actions menu on a model and choose Set default, or Clear default to return to the platform default. Agents created later without a model get the new default. Older agents still on system_model follow it immediately. Agents that name their own model, including every agent stored with the default at creation, are unaffected.

It refuses a model id, or a fallback, that the organization has not configured. That protects you: an accepted id that resolves to nothing would break every agent in the organization on its next run. To change one agent, put the id in that agent's llm.model and leave this alone; see list-models.

Every tool call works the same way: How every tool call works.