Qwen3.8 27B EXL3 3.5bpw — Hardware Requirements & GPU Compatibility
Chat
Qwen3.8 27B EXL3 3.5bpw is a 7.7B-parameter open language model from Mia-AiLab in the Qwen 3.8 family. It supports a context window of up to 262,144 tokens. At BF16 it needs about 16.09 GB of VRAM — see which GPUs and Macs can run it below.
Runs well on a 24 GB GPU
Runs on 24 GB (BF16 · ~41 tok/s · 128K)Macs from 24 GB (BF16 · ~13 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Qwen3.8 27B EXL3 3.5bpw
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve Mia-AiLab/Qwen3.8-27B-EXL3-3.5bpw --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on RTX A5000 (24 GB) (7 GB headroom) · BF16
- Generation speed
- ~31 tok/s
- generation speed
- Cost per 1M output tokens
- $1.43
- per 1M output tokens
How Much VRAM Does Qwen3.8 27B EXL3 3.5bpw Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 16.1 GB | 17.4 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Fits: 128K context · ~41 tok/s |
Selected BF16
Brain floating point 1616.5GBat 16K context
- Weights
- 15.3 GB
- KV cache
- 0.9 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 7.5 GB free · ~41 tok/s
≥50 tok/s needs ≥1.24 TB/s · RTX 5090 ~72
Which Hardware Can Run Qwen3.8 27B EXL3 3.5bpw?
BF16 · 16.1 GBQwen3.8 27B EXL3 3.5bpw (BF16) requires 16.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 21+ GB is recommended. Using the full 262K context window can add up to 13.9 GB, bringing total usage to 30.0 GB. 8 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 122 devices with unified memory can run Qwen3.8 27B EXL3 3.5bpw, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Mini M5 Pro (24 GB).
20 GB
fits · 3.5 GB free · up to 64K ctx
1 GPU
RX 7900 XT~30Simulate output on RX 7900 XT32–40 GB
fits · 15.5 GB free · up to 256K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~72Simulate output on RTX 5090M5 Max36 GB~20Simulate output on Mac Studio M5 Max (36 GB)M532 GB~7Simulate output on MacBook Air 13" M5 (32 GB)Orin~8Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~6Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at BF16
Alternatives to Qwen3.8 27B EXL3 3.5bpw
Qwen 3.8 family · 12 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| ALoDLM 8BAmazon | — | 7.6 GB | ~119 | same size |
| Ling 3.0 TinyInclusion AI | — | 5.5 GB | ~364 | same size |
| BitCPM CANN 8BOpenBMB | — | 5.6 GB | ~127 | same size |
| Gemma 4 E4BGoogle | — | 5.6 GB | ~123 | same size |
| Olmo 3 7B InstructAllen AI | — | 8.4 GB | ~114 | same size |
| Phi Mini MoE InstructMicrosoft | — | 5.4 GB | ~239 | same size |
Specifications
- Publisher
- Mia-AiLab
- Family
- Qwen 3.8
- Parameters
- 7.7B
- Architecture
- Qwen3_5ForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 248,320
- Release Date
- 2026-08-24
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Qwen3.8 27B EXL3 3.5bpw need?
Qwen3.8 27B EXL3 3.5bpw requires 16.1 GB of VRAM at BF16. Full 262K context adds up to 13.9 GB (30.0 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 7.7B × 16 bits ÷ 8 = 15.3 GB
KV Cache + Overhead ≈ 0.8 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 ≈ 14.7 GB (at full 262K context)
VRAM usage by quantization
Can I run Qwen3.8 27B EXL3 3.5bpw on a Mac?
Yes — Mac Mini M5 Pro (24 GB) and 106 other Macs can run Qwen3.8 27B EXL3 3.5bpw. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 16.1 GB of usable unified memory (RAM minus macOS overhead).
Can I run Qwen3.8 27B EXL3 3.5bpw locally?
Yes — Qwen3.8 27B EXL3 3.5bpw can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Qwen3.8 27B EXL3 3.5bpw?
At BF16, Qwen3.8 27B EXL3 3.5bpw can reach ~298 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~41 tok/s. 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 ÷ 16.1 × 0.65 = ~323 tok/s
Estimated speed at BF16 (16.1 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Qwen3.8 27B EXL3 3.5bpw?
At BF16, the download is about 15.34 GB.
Which GPUs can run Qwen3.8 27B EXL3 3.5bpw?
8 consumer GPUs can run Qwen3.8 27B EXL3 3.5bpw at BF16. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XT, AMD Radeon RX 7900 XTX. 1 GPU has plenty of headroom for comfortable inference.
Which devices can run Qwen3.8 27B EXL3 3.5bpw?
122 devices with unified memory can run Qwen3.8 27B EXL3 3.5bpw at BF16 (16.1 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, 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.