GLM 5.3 EXL3 TR3 3.25bpw — Hardware Requirements & GPU Compatibility
Chat
GLM 5.3 EXL3 TR3 3.25bpw is a 169.6B-parameter open language model from davidsyoung in the GLM 5 family. It supports a context window of up to 1,048,576 tokens. At Q4_K_M it needs about 106.00 GB of VRAM — see which GPUs and Macs can run it below.
1.1K downloads
7 likesSpecifications
- Publisher
- davidsyoung
- Family
- GLM 5
- Parameters
- 169.6B
- Architecture
- GlmMoeDsaForCausalLM
- Context Length
- 1,048,576 tokens
- Vocabulary Size
- 154,880
- Release Date
- 2026-08-28
- License
- Other
Get Started
HuggingFace
How Much VRAM Does GLM 5.3 EXL3 TR3 3.25bpw Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_Kest. | 3.40 | 76.3 GB | 2082.4 GB | 72.09 GB | 2-bit quantization with K-quant improvements |
| Q3_K_Mest. | 3.90 | 86.9 GB | 2093.0 GB | 82.69 GB | 3-bit medium quantization |
| Q4_K_Mest. | 4.80 | 106 GB | 2112.1 GB | 101.78 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_Mest. | 5.70 | 125.1 GB | 2131.2 GB | 120.86 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_Kest. | 6.60 | 144.2 GB | 2150.3 GB | 139.94 GB | 6-bit quantization, very good quality |
| Q8_0est. | 8.00 | 173.8 GB | 2180.0 GB | 169.63 GB | 8-bit quantization, near-lossless |
| BF16est. | 16.00 | 343.5 GB | 2349.6 GB | 339.26 GB | Brain floating point 16 — preferred for training |
est.= calculated VRAM estimate; no published GGUF file found for that quantization yet. Other rows are verified against real community uploads.
Which GPUs Can Run GLM 5.3 EXL3 TR3 3.25bpw?
Q4_K_M · 106 GBGLM 5.3 EXL3 TR3 3.25bpw (Q4_K_M) requires 106 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 138+ GB is recommended. Using the full 1049K context window can add up to 2006.1 GB, bringing total usage to 2112.1 GB. No single GPU has enough memory — multi-GPU or cluster setups are needed.
Which Devices Can Run GLM 5.3 EXL3 TR3 3.25bpw?
Q4_K_M · 106 GB11 devices with unified memory can run GLM 5.3 EXL3 TR3 3.25bpw, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, MacBook Pro 16" M5 Max (128 GB).
Runs great
— Plenty of headroom
Related Models
Frequently Asked Questions
- How much VRAM does GLM 5.3 EXL3 TR3 3.25bpw need?
GLM 5.3 EXL3 TR3 3.25bpw requires 106 GB of VRAM at Q4_K_M, or 343.5 GB at BF16. Full 1049K context adds up to 2006.1 GB (2112.1 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 169.6B × 4.8 bits ÷ 8 = 101.8 GB
KV Cache + Overhead ≈ 4.2 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 2010.3 GB (at full 1049K context)
VRAM usage by quantization
Q4_K_M106.0 GBQ4_K_M + full context2112.1 GB- Can NVIDIA GeForce RTX 5090 run GLM 5.3 EXL3 TR3 3.25bpw?
No — GLM 5.3 EXL3 TR3 3.25bpw requires at least 76.3 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
- What's the best quantization for GLM 5.3 EXL3 TR3 3.25bpw?
For GLM 5.3 EXL3 TR3 3.25bpw, Q4_K_M (106 GB) offers the best balance of quality and VRAM usage. Q5_K_M (125.1 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 76.3 GB.
VRAM requirement by quantization
Q2_K76.3 GBQ4_K_M ★106.0 GBQ5_K_M125.1 GBQ6_K144.2 GBQ8_0173.8 GBBF16343.5 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run GLM 5.3 EXL3 TR3 3.25bpw on a Mac?
GLM 5.3 EXL3 TR3 3.25bpw requires at least 76.3 GB at Q2_K, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run GLM 5.3 EXL3 TR3 3.25bpw locally?
Yes — GLM 5.3 EXL3 TR3 3.25bpw can run locally on consumer hardware. At Q4_K_M quantization it needs 106 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is GLM 5.3 EXL3 TR3 3.25bpw?
At Q4_K_M, GLM 5.3 EXL3 TR3 3.25bpw can reach ~45 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 106.0 × 0.65 = ~49 tok/s
Estimated speed at Q4_K_M (106 GB)
~49 tok/s~49 tok/s~45 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of GLM 5.3 EXL3 TR3 3.25bpw?
At Q4_K_M, the download is about 101.78 GB. The full-precision BF16 version is 339.26 GB. The smallest option (Q2_K) is 72.09 GB.
- Which GPUs can run GLM 5.3 EXL3 TR3 3.25bpw?
No single consumer GPU has enough VRAM to run GLM 5.3 EXL3 TR3 3.25bpw at Q4_K_M (106 GB). Multi-GPU or professional hardware is required.
- Which devices can run GLM 5.3 EXL3 TR3 3.25bpw?
18 devices with unified memory can run GLM 5.3 EXL3 TR3 3.25bpw at Q4_K_M (106 GB), including ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.