GLM 5 744B-A40B
Zhipu's GLM 5 flagship — a 744B-parameter MoE with 40B active, 128k context, and MIT licensing, built for multi-GPU server deployment and strong zh/en multilingual performance. Released February 2026.
By Zhipu AI · China
Updated 2026-09-15
When to pick this model
- Multi-GPU server deployments needing a flagship-class open MoE
- Bilingual Chinese/English chat and analysis workloads
- Long-form tasks that fit within a 128k context window
- Teams needing MIT licensing at flagship scale
VRAM requirements by quantization
| Quantization | VRAM required |
|---|---|
| Q4_K_M (recommended) | 432 GB |
| Q5_K_M | 528 GB |
| Q8_0 | 796 GB |
| FP16 (no quantization) | 1488 GB |
VRAM figures include model weights plus a typical 8k KV cache and ~600 MB runtime overhead (Ollama / llama.cpp baseline). Add headroom for higher context lengths.
In practice, GLM 5 744B-A40B is server-class even at Q4_K_M (432 GB). Stepping up to Q8_0 nearly doubles the footprint to 796 GB, and unquantized FP16 weights take 1488 GB — plan your GPU around the Q4 or Q5 figure unless you specifically need the higher fidelity.
Without a GPU, GLM 5 744B-A40B needs roughly 967 GB of system RAM to run on CPU via llama.cpp or Ollama — workable for background jobs, but far slower than GPU inference. Throughput estimates from our compatibility engine: around 1.5 tokens/sec on entry-level GPUs, on the order of 2.5 tokens/sec on a mid-range card, and up to 5 tokens/sec on high-end hardware — assuming the chosen quantization fully fits in VRAM.
What hardware do you need
The table below matches GLM 5 744B-A40B to common GPU memory tiers, using the highest-fidelity quantization that fully fits each card class. Spilling layers to system RAM works but costs most of the speed, so size your card to the quantization you actually want to run.
| GPU memory | Example cards | Best fit for GLM 5 744B-A40B |
|---|---|---|
| 8 GB | RTX 5070 Laptop, RTX 5060, RTX 5060 Ti 8GB | Does not fit — needs 432 GB at Q4_K_M |
| 12 GB | RTX 5070, RTX 5070 Ti Laptop, RTX 4080 Laptop | Does not fit — needs 432 GB at Q4_K_M |
| 16 GB | RTX 5080, RTX 4080 Super, Radeon RX 9070 XT | Does not fit — needs 432 GB at Q4_K_M |
| 24 GB | RTX 4090, Radeon RX 7900 XTX, RTX 5090 Laptop | Does not fit — needs 432 GB at Q4_K_M |
| 32 GB | RTX 5090 | Does not fit — needs 432 GB at Q4_K_M |
Which hardware should you buy to run GLM 5 744B-A40B?
To run GLM 5 744B-A40B locally at Q4, you need ~432 GB for Q4 weights alone. Hardware option to compare: BOSGAME M5 128GB / 2TB (Ryzen AI Max+ 395). This model exceeds the practical GPU memory of this mini PC. Choose a smaller model or larger infrastructure.
As an Amazon Associate, BestLLMfor earns from qualifying purchases, at no extra cost to you. It does not influence our independent rankings.
Strengths
- Native 128k context
- Efficient MoE activation (40B of 744B active per token)
- Strong bilingual zh/en performance
- Permissive MIT license
Limitations
- Requires multi-GPU server hardware (~432GB VRAM at Q4)
- Gated access on Hugging Face
- Not available for local self-hosting via Ollama outside the :cloud tag
Typical workloads
In our catalog grid, GLM 5 744B-A40B is filed under Multi-GPU Server, Bilingual zh/en, 128K Context — the use cases where its size/quality trade-off makes the most sense. Its tags translate to concrete workloads: multilingual workloads.
The 125k-token context window covers long chats and mid-sized documents, though very large retrieval workloads will need chunking. The MIT license is permissive, so shipping it inside a commercial product raises no special legal questions.
Architecture & training
Architecture: Mixture of Experts · 744B total parameters / 40B active · 128k context
Training: GLM 5 family from Zhipu AI / THUDM (Tsinghua). Server-class flagship MoE, strong multilingual zh/en.
A server-class MIT-licensed MoE flagship that excels at bilingual zh/en work, provided you have the multi-GPU hardware to run it.
Quick start
Install the runtime for your system: Windows, macOS or Linux. Check the exact model tag or GGUF quantization below; catalog IDs are not necessarily Ollama tags.
Start at 4096 tokens of context, then use ollama ps to check GPU/CPU placement. A default download may use a different quantization from the configurator’s memory estimate. Keep the free setup working before considering a kit.
# HuggingFace : THUDM/glm-5Or use the open-source MCP server to query this model from Claude Desktop, Cursor, or any MCP-compatible client.
Your private ChatGPT, free, on your own machine in an hour — LM Studio, Ollama, Open WebUI, your documents, no cloud.
- Lifetime online access
- PDF + files
- 30-day refund
Similar models worth comparing
Frequently asked questions
How much VRAM does GLM 5 744B-A40B need?
At the recommended Q4_K_M quantization, GLM 5 744B-A40B needs about 432 GB of VRAM. Q8_0 takes 796 GB, and unquantized FP16 weights take 1488 GB.
Can GLM 5 744B-A40B run without a GPU?
Yes — with roughly 967 GB of system RAM it runs CPU-only through llama.cpp or Ollama. Expect a fraction of GPU speed, which is fine for background or batch jobs but slow for interactive chat.
What context window does GLM 5 744B-A40B support?
GLM 5 744B-A40B supports a 125k-token context window (128,000 tokens).
Can I use GLM 5 744B-A40B commercially?
Yes. GLM 5 744B-A40B is released under MIT, a permissive open-source license that allows commercial use, modification and redistribution.
How fast is GLM 5 744B-A40B on consumer hardware?
Our compatibility engine estimates on the order of 2.5 tokens/sec on a mid-range GPU and up to 5 tokens/sec on high-end cards, assuming the quantization fully fits in VRAM.
Which quantization of GLM 5 744B-A40B should I download first?
Start with Q4_K_M (432 GB) — the standard size/quality sweet spot. Step up to Q5_K_M or Q8_0 only if you have VRAM headroom. It does not fit a single 24 GB consumer card — plan for multi-GPU or server hardware.