GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
AMAImedia
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
AMAImedia
Model Tree
Jackrong/Qwopus3.8-27B-Flash
Input Modalities
Output Modalities
Supported Functionality
Released as part of the NOESIS Professional Multilingual Dubbing Automation Platform (framework: DHCF-FNO — Deterministic Hybrid Control Framework for Frozen Neural Operators).
Standardized 40-shard repack of Qwopus3.8-27B-Flash for the NOESIS platform.
This is a repacked version of Qwopus3.8-27B-Flash — a Qwen3.5-based 27B parameter model optimized for flash inference with hybrid linear attention.
The repack standardizes the model to:
| Property | Value |
|---|---|
| Base model | Jackrong/Qwopus3.8-27B-Flash |
| Architecture | Qwen3.5 (Qwen3_5ForConditionalGeneration) |
| Parameters | ~27B |
| Layers | 64 |
| Shards | 40 |
| Precision | BF16 |
| Visual encoder | model-extra-00001-of-00001.safetensors (1.7 GB) |
| MTP head | model-mtp.safetensors (810 MB) |
model-00001-of-00040.safetensors through model-00040-of-00040.safetensors — text model weightsmodel-extra-00001-of-00001.safetensors — vision encodermodel-mtp.safetensors — multi-token prediction headconfig.json — model configurationtokenizer.json + tokenizer_config.json — tokenizer filesmodel.safetensors.index.json — shard indexbash
python3 -m sglang.launch_server \--model-path AMAImedia/Qwopus3.8-27B-Flash-NOESIS-BF16 \--served-model-name "Qwopus3.8-27B-Flash-NOESIS" \--tp-size 1 \--context-length 262144 \--trust-remote-code \--host 0.0.0.0 --port 8000
bibtex
@misc{qwopus3.8-flash-noesis-2026,title = {Qwopus3.8-27B-Flash — NOESIS Repack},author = {AMAImedia (Ilia Bolotnikov)},year = {2026},url = {https://huggingface.co/AMAImedia}}
Released under the Apache-2.0 license. Original model by Jackrong.