GLM-4

GLM-4 is a large language model developed by Zhipu AI, available on NVIDIA Build's API catalog as one of the notable free-tier model options. It is a strong bilingual model supporting both Chinese and English, and is described as capable of handling reasoning, instruction-following, and code tasks. [NVIDIA NIM Free Models Guide]

GLM-4 is accessible through NVIDIA NIM (NVIDIA Inference Microservices) via the hosted API catalog at build.nvidia.com. Unlike credit-based models, GLM-4 is designated as free — meaning it does not consume your NVIDIA API credit balance. [NVIDIA NIM Free Models Guide]

The model is served through NVIDIA NIM's OpenAI-Compatible API endpoint:

Base URL: https://integrate.api.nvidia.com/v1
Model ID:  zhipuai/glm-4  (check the catalog page for the exact versioned string)

Note: Always verify the exact model identifier on the catalog page. The format follows organization/model-name and may include versioned variants. Passing an incorrect model ID is the most common source of integration errors. [NVIDIA NIM Free Models Guide]

GLM-4 performs comparably to much larger models on straightforward tasks. The source identifies the following categories where it consistently delivers value: [NVIDIA NIM Free Models Guide]

For complex multi-step reasoning, nuanced writing, or specialized code generation, premium models such as GPT-4o or Claude Sonnet generally outperform GLM-4. The recommended approach is to test on your specific tasks rather than relying on general benchmarks. [NVIDIA NIM Free Models Guide]

A key pattern enabled by GLM-4's free availability is the Model Tiering Strategy — routing tasks by complexity and cost sensitivity:

This tiering approach can reduce API spend by 40–70% on typical agentic workflows without meaningful quality degradation on the tasks routed to free models. [NVIDIA NIM Free Models Guide]

GLM-4 integrates into any framework that supports the OpenAI-Compatible API. See NVIDIA NIM Framework Integrations for a full overview. Quick-reference examples are below.

OpenAI Python SDK

from openai import OpenAI

client = OpenAI(
    api_key="nvapi-xxxxxxxxxxxxxxxx",
    base_url="https://integrate.api.nvidia.com/v1"
)

response = client.chat.completions.create(
    model="zhipuai/glm-4",
    messages=[
        {"role": "user", "content": "Explain this function in one paragraph."}
    ],
    temperature=0.3,
    max_tokens=512
)

print(response.choices[0].message.content)

LangChain

from langchain_openai import ChatOpenAI

nim_llm = ChatOpenAI(
    model="zhipuai/glm-4",
    api_key="nvapi-xxxxxxxxxxxxxxxx",
    base_url="https://integrate.api.nvidia.com/v1",
    temperature=0.2
)

CrewAI

from crewai import LLM

nim_model = LLM(
    model="openai/zhipuai/glm-4",  # CrewAI prefixes with "openai/"
    api_key="nvapi-xxxxxxxxxxxxxxxx",
    base_url="https://integrate.api.nvidia.com/v1"
)

AutoGen

config_list = [
    {
        "model": "zhipuai/glm-4",
        "api_key": "nvapi-xxxxxxxxxxxxxxxx",
        "base_url": "https://integrate.api.nvidia.com/v1",
        "api_type": "openai"
    }
]

[NVIDIA NIM Free Models Guide]

• Rate limits: Free-tier models have rate limits (requests per minute and a daily cap). For batch jobs, implement queuing and retry logic with exponential backoff. [NVIDIA NIM Free Models Guide]
• Context window: Verify that GLM-4's context window can accommodate your inputs. Sending inputs that exceed the limit will fail. [NVIDIA NIM Free Models Guide]
• System prompt portability: Prompts tuned for Claude or GPT-4o may not transfer directly. Test and adjust prompts for GLM-4's behavior. [NVIDIA NIM Free Models Guide]
• Output quality differences: GLM-4 is capable but not identical to frontier models. Run representative examples through both a premium model and GLM-4 before committing to routing a task category to it. [NVIDIA NIM Free Models Guide]
• Production use: Free-tier access is subject to NVIDIA's terms of service and rate limits. For high-volume production workloads, a paid tier or self-hosted NIM deployment may be required. [NVIDIA NIM Free Models Guide]

Using GLM-4 via the NVIDIA NIM API catalog requires no NVIDIA GPU on the caller's side. The inference runs entirely on NVIDIA's hosted infrastructure. NVIDIA GPUs are only required when self-hosting NIM containers on your own infrastructure, which is a separate product from the free API access. [NVIDIA NIM Free Models Guide]

• NVIDIA NIM Free Models Guide — Full guide on connecting NIM free models to agentic tools
• NVIDIA NIM (NVIDIA Inference Microservices) — The underlying inference platform
• NVIDIA Build — The API catalog where GLM-4 is listed
• OpenAI-Compatible API — The API standard GLM-4 is served through
• Model Tiering Strategy — Pattern for mixing free and premium models
• NVIDIA NIM Framework Integrations — LangChain, CrewAI, AutoGen, and more
• MindStudio — No-code platform with 200+ models including NIM models, no API key management required
• Free vs. Paid AI Model Access: NVIDIA NIM vs. $20/month Subscriptions — Comparison of access tiers