GLM-4.7-Flash

GLM-4.7-Flash is a 30B-A3B Mixture-of-Experts (MoE) large language model belonging to the GLM-4.7 family. It is designed for efficient, lightweight deployment.

The notation 30B-A3B means the model has 30 billion total parameters, but only ~3 billion are activated per token during inference. This means it computes like a 3B model per token, keeping latency and cost low despite its large total capacity.

Optimized for coding, reasoning, agent workflows, and long-context tasks, it balances high performance similar to larger models with low compute cost, making it practical for local or small-scale deployment.

When should you use GLM-4.7 Flash?

GLM-4.7-Flash is best suited for:

• Cost-sensitive deployments requiring low per-token compute
• Long-context tasks exceeding 100K tokens (e.g., document analysis, large codebases)
• Local or small-scale inference where full 30B dense models are impractical

Supported input and output formats

Input type: Text only

Output formats: Text

Model Key: crm-di-glm47b_30b_it

Capabilities

• Strong coding performance with high SWE-bench scores relative to model size
• Multi-step reasoning stability across complex chains
• Native tool calling support for Agentic workflows and multi-step workflows

Model details

To view the model details, go to the LLM Serving, select the Models tab, and choose the GLM-4.7-Flash model.

• Model Size: The model comprises 30 billion (30B) parameters, enabling high-level language understanding and generation.
• Active Parameters/token: The model has 30 billion parameters in total, it only uses about 3 billion of them to process each word (token). This keeps it fast and efficient without sacrificing quality.
• Architecture: The Mixture-of-Experts design routes each token to a sparse subset of expert layers, enabling large total capacity while keeping per-token compute low.
• Input Context Window: Supports inputs up to 200K tokens, allowing for processing of very large contexts.
• Max Output Tokens: Will be able to generate up to 128,000 tokens as output response.

The MoE design routes each token to a subset of expert layers. In a Mixture-of-Experts model, the parameters are split into groups called “experts.” For each token, only a small number of these expert groups are selected and used. This reduces the compute per token while keeping the total model capacity large.

Model parameters

QuickML provides a robust and flexible LLM (Large Language Model) serving environment, enabling developers and businesses to fine-tune AI model behaviour through a variety of customizable parameters. Unlike other platforms that restrict user control, QuickML prioritises transparency and adaptability—allowing you to balance creativity with precision, determine the ideal response length, and guide the model to behave in ways that best serve your domain-specific needs.

The GLM 4.7 Flash model exposes a set of advanced generation parameters in the LLM Serving playground that let you control how the model behaves during the response generation. These parameters fine-tune the model’s context handling, reasoning, tool calling, and thinking, which gives you fine-grained control over response quality, creativity, and length.

Whether you’re building an internal assistant, a tool that needs external functions to access live information, or a customer support bot, QuickML’s parameter configuration options ensure your application delivers responses that are not only relevant, but also contextually tailored to your use case.

Here’s a breakdown of the available parameters:

Access the model

The GLM-4.7-Flash model can be accessed by creating an LLM Serving or RAG endpoint.

Refer to the below help documentation to understand the process of creating endpoints for LLM and RAG.

1. LLM Serving — LLM Endpoints
2. RAG endpoints — RAG Endpoints

Once the endpoint is created, API details and the relevant SDK information will be provided for integration in downstream applications.

1. API Details

The following details apply to the endpoint created for this model.

• Endpoint URL: The Model API URL used to send prompts to the model.
• OAuth Scope: The permission level required to access the model.
• Authentication: Specifies OAuth as the method for verifying user identity.
• HTTP Method: Indicates that API requests must be made using the POST method.
• Headers: Requires metadata, including organization ID and OAuth token, for authorization.
• Sample Request: A sample input JSON showing how to send a prompt to the GLM-4.7-Flash model along with parameters such as top_p, temperature, and max_tokens.

{
  "model": "glm-4.7-flash",
  "messages": [{ "role": "user", "content": "Your prompt here" }],
  "temperature": 0.7,
  "top_p": 0.9,
  "max_tokens": 1024
}

• Sample Response: The output JSON format containing the model’s generated response based on the prompt.

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "model": "glm-4.7-flash",
  "choices": [{
    "message": { "role": "assistant", "content": "Response text here" },
    "finish_reason": "stop"
  }],
  "usage": { "prompt_tokens": 12, "completion_tokens": 48, "total_tokens": 60 }
}

2. SDK Integration

In addition to calling the endpoint directly over REST, you can invoke it programmatically using the Catalyst SDKs, which handles authentication, trigger requests, and response parsing for you.

QuickML endpoints are executed through a two-step pattern: create a QuickML component instance, then call either of the converseWithLlm() or askLlm() methods, passing the endpoint key (the unique ID of the published endpoint, copied from the Catalyst console) and your input data as key-value pairs. The method returns the model’s output as a structured response.

The SDK is available across languages, refer to the relevant guide for setup, authentication configuration, and method syntax:

• JavaScript — Overview · Create QuickML Instance · Execute QuickML Endpoint
• Java — Execute QuickML Endpoints
• Python — Python SDK documentation

For JavaScript, install the @zcatalyst/quickml package and import the QuickML module before using the SDK methods. Executing an endpoint requires the Admin scope.