Open-weight model
Qwen3.8-27B-W4A8-arm-FlagOS-Express
by FlagRelease FlagRelease/Qwen3.8-27B-W4A8-arm-FlagOS-Express
Qwen3.8-27B-W4A8-arm-FlagOS-Express is an open-weight model from FlagRelease. It has 6.2B parameters and a 262,144-token context. At 16-bit it needs about 15 GB of GPU memory, which fits on 1x MI300X from $1.85 an hour, at the lowest prices in the SAVRN Index.
Alibaba open‑sources the vision‑language model Qwen3.8‑27B, and the Zhongzhi (众智) FlagOS community simultaneously completes Day 0 multi‑chip adaptation; based on the FlagOS unified open‑source technology stack, Qwen3.8‑27B has finished multi‑chip adaptation…
Runs On
What it takes to serve Qwen3.8-27B-W4A8-arm-FlagOS-Express (6.2B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them.
| Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits |
|---|---|---|---|---|---|
| 16-bit | 12.5 GB | 15.0 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 8-bit | 6.2 GB | 7.5 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
| 4-bit | 3.1 GB | 3.7 GB | 1x MI300X (192 GB) Vultr |
$1.85 | 1x H100 $1.99 · 1x MI325X $2.00 |
Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the SAVRN Index, read Oct 22, 2026.
Qwen3.8-27B-W4A8-arm-FlagOS-Express on every accelerator the SAVRN Index prices, at every precision
Model Card
Alibaba open‑sources the vision‑language model Qwen3.8‑27B, and the Zhongzhi (众智) FlagOS community simultaneously completes Day 0 multi‑chip adaptation; based on the FlagOS unified open‑source technology stack, Qwen3.8‑27B has finished multi‑chip adaptation, precision alignment and deployment verification across 11 AI chips including T-Head（平头哥）、NVIDIA（英伟达）、Moore Threads（摩尔线程）、MetaX（沐曦）、Kunlunxin（昆仑芯）、Ascend（华为昇腾）、Hygon（海光）、Iluvatar CoreX（天数智芯）、Tsingmicro（清微智能）、Enflame（燧原科技）and Sunrise（曦望）, among which NVIDIA and Moore Threads support FP8 precision deployment while the rest run on BF16, and FlagOS has for the first time extended the adaptation of the latest Qwen model to ARM edge‑side…
Excerpt from the card by FlagRelease.
Configuration
- Architecture
- Qwen3_5ForConditionalGeneration
- Context length (tokens)
- 262,144
- Layers
- 64
- Hidden size
- 5,120
- Feed-forward size
- 17,408
- Attention heads
- 24
- Key/value heads
- 4
- Head dimension
- 256
- Vocabulary size
- 248,320
- Model type
- qwen3_5
- Quantization
- compressed-tensors
Identity and Version
- Repository
- FlagRelease/Qwen3.8-27B-W4A8-arm-FlagOS-Express
- Publisher
- FlagRelease
- Task
- Not stated by the source
- Modality
- Other
- Library
- Not stated by the source
- Parameters
- 6.2B parameters
- Languages
- Not stated by the source
- Revision
- ef7a21b0134c5827021710902dae10bd5d4502d3
- First published
- 2026-10-08
- Last updated
- 2026-10-08
Files and Weights
26 files, 19.4 GB in total. The weights are 10 files totalling 19.4 GB in safetensors.
Every file
| File | Type | Size | SHA-256 |
|---|---|---|---|
| model-00001-of-00009.safetensors | Weights | 2.5 GB | 54d83c1d3663 |
| model-00002-of-00009.safetensors | Weights | 2.5 GB | 2e1bf62cbcd4 |
| model-00003-of-00009.safetensors | Weights | 2.0 GB | 2327da6f3763 |
| model-00004-of-00009.safetensors | Weights | 2.0 GB | 0feb7ad38b51 |
| model-00005-of-00009.safetensors | Weights | 2.0 GB | 7021a7902d8c |
| model-00006-of-00009.safetensors | Weights | 2.0 GB | 4d96217c772d |
| model-00007-of-00009.safetensors | Weights | 2.0 GB | ec1ef9e8c180 |
| model-00008-of-00009.safetensors | Weights | 2.0 GB | 9d80337e1518 |
| model-00009-of-00009.safetensors | Weights | 1.6 GB | 46b138553b6f |
| model-mtp.safetensors | Weights | 849.4 MB | 90fa0e3eed5a |
| config.json | Configuration | 15.0 KB | — |
| configuration.json | Configuration | 29 B | — |
| generation_config.json | Configuration | 214 B | — |
| model.safetensors.index.json | Configuration | 219.7 KB | — |
| preprocessor_config.json | Configuration | 390 B | — |
| qwen38_w4a8_gptq_packed_manifest.json | Configuration | 2.5 KB | — |
| recipe.yaml | Configuration | 295 B | — |
| video_preprocessor_config.json | Configuration | 385 B | — |
| README.md | Documentation | 11.3 KB | — |
| chat_template.jinja | Other | 9.0 KB | — |
| crc32.txt | Other | 238 B | — |
| .gitattributes | Repository | 1.6 KB | — |
| merges.txt | Tokenizer | 3.4 MB | — |
| tokenizer.json | Tokenizer | 20.0 MB | 06b9509352d2 |
| tokenizer_config.json | Tokenizer | 1.1 KB | — |
| vocab.json | Tokenizer | 6.7 MB | — |
License and Download
- License
- Not stated by the source
- Access
- Open weights, no gate
- Download size
- 19.4 GB
Released by FlagRelease through its official repository on Hugging Face.
Memory Requirements
| Precision | Weights in memory |
|---|---|
| As published | 19.4 GB |
| 16-bit | 12.5 GB |
| 8-bit | 6.2 GB |
| 4-bit | 3.1 GB |
Weights only, from the published parameter count; the key-value cache and runtime add to this.
Questions About Qwen3.8-27B-W4A8-arm-FlagOS-Express
How much GPU memory does Qwen3.8-27B-W4A8-arm-FlagOS-Express need?
About 15 GB at 16-bit and 3.7 GB at 4-bit: the weights (6.2B parameters) plus a working margin. A long context needs more.
What is the cheapest GPU to run Qwen3.8-27B-W4A8-arm-FlagOS-Express on?
At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
What is Qwen3.8-27B-W4A8-arm-FlagOS-Express's context length?
262,144 tokens, from the maximum position embeddings in its published configuration.