Skip to content

How to Use GLM Through OpenRouter and Alternative Providers

Purpose

I want to use GLM models, but Zhipu’s direct service has been unreliable during training cycles. This post shows how to access GLM through alternative providers like OpenRouter.

Why Use Alternative Providers?

The problem with Zhipu’s direct service:

Issues with direct access
- Long context degraded during training cycles
- Service outages without clear communication
- Annual subscribers locked into unreliable service
- Quantized models served without notification

The solution: Use alternative providers that offer GLM model access.

A Reddit user confirmed this works: “GLM 5 through other providers seems to work fine. Even openrouter I find pretty good.”

Option 1: OpenRouter

OpenRouter is a unified API that aggregates multiple AI models from different providers.

How to set it up:

  1. Create an account at openrouter.ai
  2. Add API credits (pay per token)
  3. Use the OpenRouter API endpoint

API call example:

openrouter_glm.py
import openai
# Configure OpenRouter as base URL
client = openai.OpenAI(
base_url="https://openrouter.ai/api/v1",
api_key="YOUR_OPENROUTER_API_KEY"
)
# Use GLM model
response = client.chat.completions.create(
model="zhipu/glm-4-plus",
messages=[
{"role": "user", "content": "Hello, how can you help me?"}
]
)
print(response.choices[0].message.content)

Key points:

  • The model name format is zhipu/glm-4-plus (provider prefix)
  • Uses standard OpenAI SDK interface
  • Pay only for what you use
  • No annual commitment

Option 2: Alibaba Coding Plan

Alibaba’s coding platform includes GLM access as part of their subscription.

A user reported: “I’m using it through alibaba coding plan and its great.”

Benefits:

  • Bundled with Alibaba Cloud services
  • Optimized for coding tasks
  • Subscription-based (predictable costs)
  • Good integration with Alibaba ecosystem

Option 3: Self-Hosting

GLM is open-source, so you can run it yourself.

Requirements:

Self-hosting requirements
- Download weights from HuggingFace
- GPU with significant VRAM (for large models)
- Technical skill to deploy and maintain
- No API costs, but infrastructure costs

This gives you full control and privacy, but requires hardware investment.

Provider Comparison

Provider comparison
Provider Pricing Flexibility Best For
─────────────────────────────────────────────────────
Zhipu Direct Subscription Limited Official support
OpenRouter Pay per token High Testing, flexibility
Alibaba Subscription Medium Coding workflows
Self-host Hardware cost Full Privacy, high volume

Common Issues

Issue: Model names differ between providers

The same model might be called:

  • glm-4 (Zhipu direct)
  • zhipu/glm-4-plus (OpenRouter)
  • Different naming in Alibaba

Solution: Check the provider’s documentation for exact model identifiers.

Issue: Different context limits

Some providers may limit context differently than Zhipu direct.

Solution: Verify the context window with your chosen provider before committing.

Issue: Feature availability

Not all providers support all GLM variants or features (function calling, vision).

Solution: Confirm specific features are supported before building dependencies.

Using with LangChain

If you use LangChain, integrating OpenRouter is straightforward:

langchain_openrouter.py
from langchain_openai import ChatOpenAI
# OpenRouter with LangChain
llm = ChatOpenAI(
model="zhipu/glm-4-plus",
openai_api_base="https://openrouter.ai/api/v1",
openai_api_key="YOUR_OPENROUTER_API_KEY"
)
response = llm.invoke("What is the capital of France?")
print(response.content)

This approach works with any LangChain-based application.

Summary

In this post, I showed how to access GLM through alternative providers. The key point is that OpenRouter and other providers let you use GLM models even when Zhipu’s direct service has issues.

The main options are:

  • OpenRouter for pay-per-token flexibility
  • Alibaba Coding Plan for coding-focused subscriptions
  • Self-hosting for full control

Using alternative providers gives you redundancy and flexibility that annual subscriptions don’t offer.

Final Words + More Resources

My intention with this article was to help others share my knowledge and experience. If you want to contact me, you can contact by email: Email me

Here are also the most important links from this article along with some further resources that will help you in this scope:

Oh, and if you found these resources useful, don’t forget to support me by starring the repo on GitHub!

Comments