GLM 5.3 UNCENSORED EXL3 3.0bpw — Hardware Requirements & GPU Compatibility
ChatFunctions
GLM 5.3 UNCENSORED EXL3 3.0bpw is a 146.3B-parameter open language model from Infatoshi in the GLM 5 family. It supports a context window of up to 1,048,576 tokens. At Q4_K_M it needs about 92.01 GB of VRAM — see which GPUs and Macs can run it below.
2.3K downloads
309 likesSpecifications
- Publisher
- Infatoshi
- Family
- GLM 5
- Parameters
- 146.3B
- Architecture
- GlmMoeDsaForCausalLM
- Context Length
- 1,048,576 tokens
- Vocabulary Size
- 154,880
- Release Date
- 2026-10-01
- License
- MIT
Get Started
HuggingFace
Run in cloud
Fits on H200 (141 GB) (48 GB headroom) · Q4_K_M
- Generation speed
- ~34 tok/s
- generation speed
- Cost per 1M output tokens
- $29.42
- per 1M output tokens
How Much VRAM Does GLM 5.3 UNCENSORED EXL3 3.0bpw Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_Kest. | 3.40 | 66.4 GB | 156.7 GB | 62.18 GB | 2-bit quantization with K-quant improvements |
| Q3_K_Mest. | 3.90 | 75.5 GB | 165.8 GB | 71.33 GB | 3-bit medium quantization |
| Q4_K_Mest. | 4.80 | 92.0 GB | 182.3 GB | 87.79 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_Mest. | 5.70 | 108.5 GB | 198.8 GB | 104.25 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_Kest. | 6.60 | 124.9 GB | 215.2 GB | 120.71 GB | 6-bit quantization, very good quality |
| Q8_0est. | 8.00 | 150.5 GB | 240.8 GB | 146.31 GB | 8-bit quantization, near-lossless |
| BF16est. | 16.00 | 296.9 GB | 387.1 GB | 292.62 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 UNCENSORED EXL3 3.0bpw?
Q4_K_M · 92.0 GBGLM 5.3 UNCENSORED EXL3 3.0bpw (Q4_K_M) requires 92.0 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 120+ GB is recommended. Using the full 1049K context window can add up to 90.3 GB, bringing total usage to 182.3 GB. No consumer GPU has enough memory.
Which Devices Can Run GLM 5.3 UNCENSORED EXL3 3.0bpw?
Q4_K_M · 92.0 GB18 devices with unified memory can run GLM 5.3 UNCENSORED EXL3 3.0bpw, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, MacBook Pro 16" M5 Max (128 GB).
Runs great
— Plenty of headroom
Decent
— Enough memory, may be tightRelated Models
Frequently Asked Questions
- How much VRAM does GLM 5.3 UNCENSORED EXL3 3.0bpw need?
GLM 5.3 UNCENSORED EXL3 3.0bpw requires 92.0 GB of VRAM at Q4_K_M, or 296.9 GB at BF16. Full 1049K context adds up to 90.3 GB (182.3 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 146.3B × 4.8 bits ÷ 8 = 87.8 GB
KV Cache + Overhead ≈ 4.2 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 94.5 GB (at full 1049K context)
VRAM usage by quantization
Q4_K_M92.0 GBQ4_K_M + full context182.3 GB- Can NVIDIA GeForce RTX 5090 run GLM 5.3 UNCENSORED EXL3 3.0bpw?
No — GLM 5.3 UNCENSORED EXL3 3.0bpw requires at least 66.4 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 UNCENSORED EXL3 3.0bpw?
For GLM 5.3 UNCENSORED EXL3 3.0bpw, Q4_K_M (92.0 GB) offers the best balance of quality and VRAM usage. Q5_K_M (108.5 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 66.4 GB.
VRAM requirement by quantization
Q2_K66.4 GBQ4_K_M ★92.0 GBQ5_K_M108.5 GBQ6_K124.9 GBQ8_0150.5 GBBF16296.9 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run GLM 5.3 UNCENSORED EXL3 3.0bpw on a Mac?
Yes, but only at lower quantizations. The smallest Mac that can run GLM 5.3 UNCENSORED EXL3 3.0bpw is Mac Studio (M3 Ultra, 96GB) at Q2_K; 7 of the 39 Macs we list can run it at some quantization. For Q4_K_M (92.0 GB) you need a Mac with more unified memory.
- Can I run GLM 5.3 UNCENSORED EXL3 3.0bpw locally?
Yes — GLM 5.3 UNCENSORED EXL3 3.0bpw can run locally on consumer hardware. At Q4_K_M quantization it needs 92.0 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is GLM 5.3 UNCENSORED EXL3 3.0bpw?
At Q4_K_M, GLM 5.3 UNCENSORED EXL3 3.0bpw can reach ~52 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 ÷ 92.0 × 0.65 = ~57 tok/s
Estimated speed at Q4_K_M (92.0 GB)
~57 tok/s~57 tok/s~52 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 UNCENSORED EXL3 3.0bpw?
At Q4_K_M, the download is about 87.79 GB. The full-precision BF16 version is 292.62 GB. The smallest option (Q2_K) is 62.18 GB.
- Which GPUs can run GLM 5.3 UNCENSORED EXL3 3.0bpw?
No single consumer GPU has enough VRAM to run GLM 5.3 UNCENSORED EXL3 3.0bpw at Q4_K_M (92.0 GB). Multi-GPU or professional hardware is required.
- Which devices can run GLM 5.3 UNCENSORED EXL3 3.0bpw?
18 devices with unified memory can run GLM 5.3 UNCENSORED EXL3 3.0bpw at Q4_K_M (92.0 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.