How do I use GLM 5.3 in OpenClaw?
Add z-ai/glm-5.3 and z-ai/glm-5.3-flash to a standardcompute provider in ~/.openclaw/openclaw.json (baseUrl https://api.stdcmpt.com/v1), make standardcompute/z-ai/glm-5.3 the primary model and standardcompute/z-ai/glm-5.3-flash the imageModel: OpenClaw runs GLM 5.3 with no Z.ai account, on a plan from $19 a month that answers HTTP 402 to new requests once its compute budget is used.
{
"models": {
"mode": "merge",
"providers": {
"standardcompute": {
"baseUrl": "https://api.stdcmpt.com/v1",
"apiKey": "${STANDARDCOMPUTE_API_KEY}",
"api": "openai-responses",
"models": [
{
"id": "z-ai/glm-5.3",
"name": "GLM 5.3",
"reasoning": true,
"input": ["text"],
"contextWindow": 400000,
"maxTokens": 32000
},
{
"id": "z-ai/glm-5.3-flash",
"name": "GLM 5.3 Flash",
"reasoning": true,
"input": ["text", "image"],
"contextWindow": 400000,
"maxTokens": 32000
}
]
}
}
},
"agents": {
"defaults": {
"model": { "primary": "standardcompute/z-ai/glm-5.3" },
"imageModel": { "primary": "standardcompute/z-ai/glm-5.3-flash" }
}
}
}
Every model an agent refers to needs its own entry in the models list, imageModel included, or OpenClaw cannot resolve the ref.
GLM 5.3 needs an image model
GLM 5.3 is text-only, so its entry says "input": ["text"] and OpenClaw does not inject images into its turns. imageModel is what the view_image tool uses when the primary cannot accept images; pointing it at GLM 5.3 Flash, marked text and image, keeps screenshots and photos sent over chat working.
What a day of GLM costs
| Per day | GLM 5.3 | GLM 5.3 Flash |
|---|---|---|
| Typical day | $3.65 | $0.41 |
| Busy day | $37 | $4.14 |
| Typical days in the Starter plan's $20 compute budget | 5 | 49 |
Z.ai list prices per million tokens: GLM 5.3 $1.40 input, $0.26 cached, $4.40 output; Flash $0.15, $0.03, $0.50. Source. The Agent Burn Index measures coding agents such as OpenCode, Cline and Kilo; OpenClaw is not broken out, so these figures use its median and 90th-percentile coding-agent day (September 13 to October 12, 2026) as a stand-in. Priced at the list rates above; your own days will differ.
What else to expect
- GLM always reasons.
/think offomits the effort, so GLM reasons at its default; an explicit none would be raised to low. - A heartbeat model can differ from the primary: GLM 5.3 Flash costs about a ninth as much per day, which suits the every-30-minutes check-ins.
The catch
OpenClaw's Z.ai plugin (@openclaw/zai-provider, ZAI_API_KEY) serves zai/glm-5.3, which its docs call a Coding Plan model, so a GLM Coding Plan from $18 a month is the direct route if GLM is all you run.
Where Standard Compute fits
GLM 5.3 as primary, Flash as imageModel and Claude Sonnet 5.5 in fallbacks is one provider block and one key, with a monthly cap on all three.
FAQ
Is the OpenClaw model ref zai/glm-5.3 or standardcompute/z-ai/glm-5.3?
zai/glm-5.3 is OpenClaw's Z.ai plugin on a Z.ai key. standardcompute/z-ai/glm-5.3 is the same model through this config; the ref splits on the first slash.
Does adding this provider remove OpenClaw's built-in providers?
No. "mode": "merge" keeps them next to standardcompute; "replace" would leave only the providers you configure.