GLM (Zhipu AI) with OpenClaw

Configure GLM-5, GLM-4.7 via Volcano Engine or Z.AI

GLM (Zhipu AI) models are supported via Volcano Engine (volcengine) or Z.AI (zai). Volcano Engine offers GLM-4.7; Z.AI provides GLM-5, GLM-4.7, and Coding Plan variants.

Volcano Engine

Use volcengine/glm-4-7-251222. See Volcano Engine for setup.

Z.AI

Onboard with openclaw onboard --auth-choice zai-api-key. Configure as zai/glm-5. Get API key from Z.AI console.

Model IDs (examples)

  • volcengine/glm-4-7-251222 — Via Volcano Engine
  • zai/glm-5, zai/glm-4-7 — Via Z.AI

Exact slugs change with provider releases—confirm in openclaw models list after onboard.

When to pick GLM

  • Chinese and bilingual tasks — Strong regional language support.
  • Coding Plan variants — Cheaper agent loops for repo work when paired with tool limits.
  • APAC latency — Often better than routing everything through US-only APIs.

For tiered routing (cheap model for heartbeats, flagship for chat), see Example setups & model routing.

Choosing Volcano vs Z.AI

Volcano Engine is a single cloud path with dated model IDs (e.g. glm-4-7-251222). Z.AI exposes newer GLM-5 and Coding Plan SKUs with zai-api-key onboard. You can configure one path per agent or split: flagship chat on Z.AI, batch tasks on Volcano if pricing differs in your region.

Onboard paths

Volcano: volcengine provider + dated model ID. Z.AI: openclaw onboard --auth-choice zai-api-key. Do not paste the same key into both consoles—billing and rate limits are separate accounts.

Related