Back to list

QWEN3.8 27

Alibaba Cloud
Qwen3.8_27b-mtp_256k_b1_2core_1chip
QWEN3 Qwen3.8 27 gguf w4a8 262,144 tokens M50 Stable 2.1.0
Download Model
https://dr.bee-link.cn/vfm-admin/vfm-downloader.php?q=dXBsb2Fkcy9MTE0vTW9kZWxzL1FXRU4zL1F3ZW4zLjgvSGlNb2RlbC1Rd2VuMy44LTI3Yi1tdHAtdzRhOC1wMjU2LTI1NkstYjEtMmNvcmUtMWQteGgyLXYxLjUuMC0yMDI2MDgxODE3NDgzMS5nZ3Vm&h=36b8636774b6431ba815155f1b1be116
Download Now
SHA-256 Checksum 21.15 GB
bd12bd29c17feba7358c8e9f5f887a342dd5c94cae509b23081ce7e032455f33
Verify the download: on Windows run certutil -hashfile FILENAME SHA256; on macOS / Linux run shasum -a 256 FILENAME or sha256sum FILENAME. A matching value means the file is intact.
Build: 1.5.0
Note: Models built on v1.3 require NPU driver 1.3 or above to work properly.
Note: Models for inference framework 2.1.0 require inference framework 2.1.0 or above to work properly.
27
Parameters
262,144
Tokens
gguf
Format
—
Target
1
Batch
2
CPU Cores

Basic Info

Model ID Qwen3.8_27b-mtp_256k_b1_2core_1chip
Vendor QWEN3.8
Name (CN) QWEN3.8 27
Name (EN) QWEN3.8 27
Series QWEN3
Type Qwen3.8

Tech Specs

Format gguf
Parameters 27
Quantization w4a8
Context Length 262144
Batch ✓ true
CPU Cores 2
Chip Count ✓ true
Chip Type M50

Deployment

Available ✓ true
Status Stable
Software Stack 2.1.0

Capabilities

Input Text
Output Text
Features gguf w4a8
Inference Engines HLIEvLLM HLIELLama
Frameworks HLIEvLLM HLIELLama

Description

Description 需要24G+的卡
Description (CN) 需要24G+的卡

Build Info

Build Version 1.5.0

Timestamps

Created 2026-11-07 06:00:02
Updated 2026-11-07 06:00:02