Keys GLM 5.3 EXL3 Abliterated — Hardware Requirements & GPU Compatibility
Chat
Keys GLM 5.3 EXL3 Abliterated is a 165.0B-parameter open language model from drowzeys in the GLM 5 family. At Q4_K_M it needs about 108.92 GB of VRAM — see which GPUs and Macs can run it below.
69 downloads
4 likesSpecifications
- Publisher
- drowzeys
- Family
- GLM 5
- Parameters
- 165.0B
- Release Date
- 2026-09-04
- License
- MIT
Get Started
HuggingFace
How Much VRAM Does Keys GLM 5.3 EXL3 Abliterated Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_Kest. | 3.40 | 77.2 GB | — | 70.14 GB | 2-bit quantization with K-quant improvements |
| Q3_K_Mest. | 3.90 | 88.5 GB | — | 80.45 GB | 3-bit medium quantization |
| Q4_K_Mest. | 4.80 | 108.9 GB | — | 99.02 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_Mest. | 5.70 | 129.3 GB | — | 117.58 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_Kest. | 6.60 | 149.8 GB | — | 136.15 GB | 6-bit quantization, very good quality |
| Q8_0est. | 8.00 | 181.5 GB | — | 165.03 GB | 8-bit quantization, near-lossless |
| BF16est. | 16.00 | 363.1 GB | — | 330.06 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 Keys GLM 5.3 EXL3 Abliterated?
Q4_K_M · 108.9 GBKeys GLM 5.3 EXL3 Abliterated (Q4_K_M) requires 108.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 142+ GB is recommended. No single GPU has enough memory — multi-GPU or cluster setups are needed.
Which Devices Can Run Keys GLM 5.3 EXL3 Abliterated?
Q4_K_M · 108.9 GB11 devices with unified memory can run Keys GLM 5.3 EXL3 Abliterated, 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 Keys GLM 5.3 EXL3 Abliterated need?
Keys GLM 5.3 EXL3 Abliterated requires 108.9 GB of VRAM at Q4_K_M, or 363.1 GB at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 165.0B × 4.8 bits ÷ 8 = 99 GB
KV Cache + Overhead ≈ 9.9 GB (at 2K context + ~0.3 GB framework)
VRAM usage by quantization
Q4_K_M108.9 GB- Can NVIDIA GeForce RTX 5090 run Keys GLM 5.3 EXL3 Abliterated?
No — Keys GLM 5.3 EXL3 Abliterated requires at least 77.2 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
- What's the best quantization for Keys GLM 5.3 EXL3 Abliterated?
For Keys GLM 5.3 EXL3 Abliterated, Q4_K_M (108.9 GB) offers the best balance of quality and VRAM usage. Q5_K_M (129.3 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 77.2 GB.
VRAM requirement by quantization
Q2_K77.2 GBQ4_K_M ★108.9 GBQ5_K_M129.3 GBQ6_K149.8 GBQ8_0181.5 GBBF16363.1 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run Keys GLM 5.3 EXL3 Abliterated on a Mac?
Keys GLM 5.3 EXL3 Abliterated requires at least 77.2 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 Keys GLM 5.3 EXL3 Abliterated locally?
Yes — Keys GLM 5.3 EXL3 Abliterated can run locally on consumer hardware. At Q4_K_M quantization it needs 108.9 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Keys GLM 5.3 EXL3 Abliterated?
At Q4_K_M, Keys GLM 5.3 EXL3 Abliterated can reach ~44 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 ÷ 108.9 × 0.65 = ~48 tok/s
Estimated speed at Q4_K_M (108.9 GB)
~48 tok/s~48 tok/s~44 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Keys GLM 5.3 EXL3 Abliterated?
At Q4_K_M, the download is about 99.02 GB. The full-precision BF16 version is 330.06 GB. The smallest option (Q2_K) is 70.14 GB.
- Which GPUs can run Keys GLM 5.3 EXL3 Abliterated?
No single consumer GPU has enough VRAM to run Keys GLM 5.3 EXL3 Abliterated at Q4_K_M (108.9 GB). Multi-GPU or professional hardware is required.
- Which devices can run Keys GLM 5.3 EXL3 Abliterated?
18 devices with unified memory can run Keys GLM 5.3 EXL3 Abliterated at Q4_K_M (108.9 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.