GLM-4.7

What is GLM-4.7?

GLM-4.7 is Z.ai's MoE open-weights language model. Released in 2025-12, it ships in 1 size (mid-size MoE) and is licensed MIT. The most popular use case is advanced coding and agentic workflows.

VRAM and hardware

The mid-size MoE size fits in 64 GB of VRAM.

Size	Min VRAM (Q4_K_M)	Recommended VRAM	Tokens/sec on 3090
mid-size MoE	64 GB	96 GB	~13 tok/s

How to run GLM-4.7 locally

Option 1: Ollama (simplest)

ollama pull glm-4-7
ollama run glm-4-7

Option 2: Mullama (production)

mullama pull glm-4-7
mullama run glm-4-7

Option 3: llama.cpp (CLI)

# Download a GGUF from Hugging Face (search "glm-4-7 gguf")
./llama-cli -m glm-4-7.Q4_K_M.gguf -p "Hello, AI!"

Option 4: Python with Mullama or llama-cpp-python

from mullama import Model, Context
model = Model.load("glm-4-7.Q4_K_M.gguf", n_gpu_layers=99)
ctx = Context(model, n_ctx=4096)
print(ctx.generate("Hello, AI!", 256))

What you can build with GLM-4.7

• coding
• agentic
• tool use

Hardware it fits on

• Apple Silicon (8GB): mid-size MoE in MLX
• Apple Silicon (16GB): mid-size MoE in MLX
• Apple Silicon (32GB): mid-size MoE in MLX or GGUF
• Apple Silicon (64GB+): mid-size MoE in MLX
• RTX 3090 (24GB): mid-size MoE at Q4_K_M
• RTX 4090 (24GB): same as 3090, ~30% faster
• RTX 5090 (32GB): mid-size MoE at Q8_0
• 2× RTX 6000 Ada (96GB total): mid-size MoE at Q4_K_M

Related models in the GLM family

• GLM-5 — 744B (40B active MoE) · frontier-tier reasoning and agentic engineering
• GLM-5.1 — 744B (40B active MoE) · next-generation frontier reasoning, agentic engineering
• GLM-5.2 — 744B (40B active MoE) · latest Z.ai frontier reasoning model (2026 mid-year)
• GLM-4.7 Flash — 30B · strong 30B coder that runs on a single 3090/4090

Other models you might consider

• Llama 3.1 (Meta) — 8B, general-purpose chat, agentic workflows, code
• Llama 3.2 (Meta) — 1B, low-end hardware, mobile, edge deployment
• Llama 3.2 Vision (Meta) — 11B, vision + chat on desktop GPUs
• Llama 3.3 (Meta) — 70B, best-in-class 70B for desktops
• Llama 4 (Meta) — Scout 17B (109B MoE), vision + long context (10M tokens)
• Qwen 2.5 (Alibaba) — 0.5B, strongest 7B-32B on consumer hardware