How to Use GLM-5.2 with Claude Code, Cursor, and Cline
GLM-5.2 in Claude Code takes three environment variables. This guide walks through the exact setup for Claude Code, Cursor, and Cline via LLM Gateway — plus why GLM-5.2 is the rare open coder that's uncapped on a flat-rate DevPass plan.
GLM-5.2 has quietly become the workhorse of the open-coder wave: Zhipu's flagship for long-horizon agentic engineering, a 1M-token context, reasoning support, and output listed at $4.40 per million tokens (Z.ai's list rate — several gateway providers serve it for less) — a fraction of frontier rates. It anchors Z.ai's own coding plan and sits in the uncapped tier of OpenCode Go's catalog. But Zhipu doesn't ship a coding agent, and your coding agent doesn't ship GLM-5.2.
LLM Gateway bridges that gap. It speaks both the Anthropic and OpenAI API formats, so the tools you already use can run GLM-5.2 — or any of 200+ models — with a base-URL change. Here is the exact setup for each tool.
GLM-5.2 in Claude Code
Claude Code talks to any endpoint that speaks Anthropic's /v1/messages format, which LLM Gateway does natively. Three environment variables:
1export ANTHROPIC_BASE_URL=https://api.llmgateway.io2export ANTHROPIC_AUTH_TOKEN=$LLM_GATEWAY_API_KEY3export ANTHROPIC_MODEL=glm-5.24
5claude1export ANTHROPIC_BASE_URL=https://api.llmgateway.io2export ANTHROPIC_AUTH_TOKEN=$LLM_GATEWAY_API_KEY3export ANTHROPIC_MODEL=glm-5.24
5claudeThat's the whole migration. Every primary request now routes through LLM Gateway to GLM-5.2, and every request shows up in your dashboard with its exact cost, token counts, and cache-hit rate.
One refinement worth adding: Claude Code uses a second, smaller model for routine background work, and you can point it at something nearly free:
1export ANTHROPIC_DEFAULT_HAIKU_MODEL=mimo-v2.51export ANTHROPIC_DEFAULT_HAIKU_MODEL=mimo-v2.5That puts GLM-5.2 on the real work and a model priced from $0.14/M input and $0.28/M output on the housekeeping.
GLM-5.2 in Cursor
Cursor routes its AI panel (Cmd/Ctrl + L) — both plan mode and agent mode — through a custom OpenAI-compatible endpoint. Setup:
- Open Cursor Settings → Models
- Add your LLM Gateway key under OpenAI API Key
- Enable Override OpenAI Base URL and set it to
https://api.llmgateway.io/v1 - Add
glm-5.2as a custom model and select it
Be aware of the boundary: Cursor's inline edit (Cmd/Ctrl + K) and tab autocomplete are locked to Cursor's own backend and will not route through any external endpoint. Plan, chat, and run agent tasks on GLM-5.2's full 1M context in Cursor; for a complete agent loop on it, use Claude Code or Cline instead.
Every frontier model. One flat rate.
Run Claude, GPT-5.5, Gemini, and 200+ more through Claude Code, Cursor, or whatever tool you already use. From $29/mo — no token math, no surprise invoices.
GLM-5.2 in Cline
Cline is the straightforward one — it's built to bring your own key:
- Open the Cline panel in VS Code and click the settings gear
- Set API Provider to OpenAI Compatible
- Base URL:
https://api.llmgateway.io/v1 - API Key: your LLM Gateway key
- Model ID:
glm-5.2
Cline's full agent loop — file edits, terminal commands, project scaffolding — now runs on GLM-5.2. If you want Zhipu's own deployment specifically, use zai/glm-5.2; otherwise the gateway picks a healthy provider from the dozen that serve the model and fails over automatically.
Also worth knowing: OpenCode ships LLM Gateway as a built-in provider, so there GLM-5.2 is a login and a model pick, no URLs at all.
What GLM-5.2 costs in your coding agent
Agent loops are token-hungry, which is exactly the case DevPass was built for — a flat monthly rate instead of a per-token bill:
| Plan | Price | Model usage included |
|---|---|---|
| Lite | $29/mo | ~$87 at provider rates |
| Pro | $79/mo | ~$237 |
| Max | $179/mo | ~$537 |
Here's what makes GLM-5.2 special on DevPass: it's a standard-tier model, so unlike Kimi K3 or the frontier flagships it has no weekly allowance cap — it draws only from your monthly credit pool. At GLM-5.2's list rates ($1.40/M input, $0.26/M cached input, $4.40/M output — discounted providers only improve the math), a heavy agent session — a few million tokens in, most of them cache hits — lands around $2. The Pro allowance covers over a hundred of those a month, and the cached-input rate is what makes the math work: agent loops re-send the same context every step, and cache hits bill at less than a fifth of the fresh rate.
The pattern most developers settle into: GLM-5.2 for the bulk of the loop, and a premium model — Kimi K3, Claude Opus 5 — for the planning passes and the bugs that resist it. Same key, same session, one model switch.
Prefer straight metering? Pay-as-you-go credits work with the identical setup: top up from $10 and pay the published rates plus a 5% platform fee when you buy credits (plus 1.5% for international cards).
Getting started
- Get DevPass — flat-rate GLM-5.2 in your coding agent from $29/mo, uncapped within your allowance
- Try LLM Gateway free — one key for GLM-5.2 and 200+ models
- Weighing plans? See where GLM-anchored plans rank in the best AI coding plans in 2026
- Requests failing? Check the live GLM-5.2 status page for per-provider uptime before you debug your setup
Every frontier model. One flat rate.
Run Claude, GPT-5.5, Gemini, and 200+ more through Claude Code, Cursor, or whatever tool you already use. From $29/mo — no token math, no surprise invoices.
Frequently asked questions
- Does Claude Code work with GLM-5.2?
- Yes. Claude Code sends Anthropic-format requests to whatever
ANTHROPIC_BASE_URLpoints at. LLM Gateway accepts that format and translates it for each provider, soANTHROPIC_MODEL=glm-5.2just works — with the model's full 1M-token context and reasoning support. - Is GLM-5.2 a premium-tier model on DevPass?
- No — and that's the point. Premium tiering starts at $15 per million output tokens; GLM-5.2's $4.40/M output keeps it standard tier, so it has no weekly allowance cap. It draws only from your monthly credit pool, which makes it the natural bulk-loop model on any DevPass plan.
- GLM-5.2 or Kimi K3 for coding?
- Both carry a 1M context. Kimi K3 is the stronger pick for the hardest reasoning problems but costs $15/M output and sits in DevPass's capped premium tier. GLM-5.2 is roughly 3.4× cheaper on output and uncapped weekly. The pattern most developers land on: GLM-5.2 for the bulk of the agent loop, K3 or a frontier model for the problems that resist it.
- How is this different from the Z.ai GLM Coding Plan?
- Z.ai's own plan is the cheapest way to run GLM models exclusively — one vendor, one model family. Running GLM-5.2 through LLM Gateway costs more but puts the same model next to 200+ others under one key, with automatic failover across a dozen upstream providers when one has a bad day.