Zhipu

GLM-5.2

Zhipu GLM-5.2, an MLA+DSA MoE with MTP; served at NVFP4 on B300.

744B (MoE, ~40B active)MIT

Context

Standard tok/s

TTFT

300ms

Ultra-fast tier serves at 1,000+ tps on reserved capacity, see pricing.

Pricing

Input: $1.20 / 1M tokens
Output: $3.00 / 1M tokens
Cached input: $0.2 / 1M tokens
Fine-tunes: Same per-token price as base

Capabilities

SSE streaming
Tool / function calling
Structured JSON outputs
Reasoning traces

Use cases

CodingAgenticReasoningTool use

Quickstart

Full quickstart

glm-5.2-nvfp4.py

from openai import OpenAI

client = OpenAI(
    base_url="https://api.cogito.decart.ai/v1",
    api_key=os.environ["COGITO_API_KEY"],
)

response = client.chat.completions.create(
    model="zai-org/GLM-5.2-NVFP4",
    messages=[{"role": "user", "content": "Hello!"}],
)

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