What are the hardware requirements for Qwen3.8 2.4T A95B NVFP4 REAP 25?

Qwen3.8 2.4T A95B NVFP4 REAP 25 requires approximately 899.5 GB of VRAM and supports a context window of 256K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Qwen3.8 2.4T A95B NVFP4 REAP 25 and how large is it?

Qwen3.8 2.4T A95B NVFP4 REAP 25 is developed by RedHatAI, a model with 1853.4b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate Qwen3.8 2.4T A95B NVFP4 REAP 25?

Qwen3.8 2.4T A95B NVFP4 REAP 25 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

Qwen3.8 2.4T A95B NVFP4 REAP 25 by RedHatAI — VRAM 899.5GB, 256K context

Name: Qwen3.8 2.4T A95B NVFP4 REAP 25
Author: RedHatAI

Qwen3.8 2.4T A95B NVFP4 REAP 25 is an open-source language model by RedHatAI. Features: 1853.4b LLM, VRAM: 899.5GB, Context: 256K, License: mit, MoE, LLM Explorer Score: 0.34.

8-bit Base model:quantized:qwen/qwen... Base model:qwen/qwen3.8-2.4t-a... Compressed-tensors Conversational Endpoints compatible Llm-compressor Moe Qwen3 5 moe text Region:us Safetensors Sharded Tensorflow Vllm

Model Card on HF 🤗: https://huggingface.co/RedHatAI/Qwen3.8-2.4T-A95B-NVFP4-REAP-25

Qwen3.8 2.4T A95B NVFP4 REAP 25 Parameters and Internals

LLM Name	Qwen3.8 2.4T A95B NVFP4 REAP 25
Repository 🤗	https://huggingface.co/RedHatAI/Qwen3.8-2.4T-A95B-NVFP4-REAP-25
Base Model(s)	Qwen/Qwen3.8-2.4T-A95B Qwen/Qwen3.8-2.4T-A95B
Model Size	1853.4b
Required VRAM	899.5 GB
Updated	2026-08-21
Maintainer	RedHatAI
Model Type	qwen3_5_moe_text
Model Files	20.0 GB: 1-of-55 20.0 GB: 2-of-55 20.0 GB: 3-of-55 20.0 GB: 4-of-55 20.0 GB: 5-of-55 20.0 GB: 6-of-55 20.0 GB: 7-of-55 20.0 GB: 8-of-55 20.0 GB: 9-of-55 20.0 GB: 10-of-55 20.0 GB: 11-of-55 19.8 GB: 12-of-55 20.0 GB: 13-of-55 20.0 GB: 14-of-55 20.0 GB: 15-of-55 20.0 GB: 16-of-55 20.0 GB: 17-of-55 20.0 GB: 18-of-55 20.0 GB: 19-of-55 20.0 GB: 20-of-55 20.0 GB: 21-of-55 19.9 GB: 22-of-55 20.0 GB: 23-of-55 20.0 GB: 24-of-55 20.0 GB: 25-of-55 20.0 GB: 26-of-55 20.0 GB: 27-of-55 20.0 GB: 28-of-55 20.0 GB: 29-of-55 20.0 GB: 30-of-55 20.0 GB: 31-of-55 20.0 GB: 32-of-55 20.0 GB: 33-of-55 20.0 GB: 34-of-55 20.0 GB: 35-of-55 20.0 GB: 36-of-55 20.0 GB: 37-of-55 20.0 GB: 38-of-55 20.0 GB: 39-of-55 20.0 GB: 40-of-55 20.0 GB: 41-of-55 20.0 GB: 42-of-55 20.0 GB: 43-of-55 20.0 GB: 44-of-55 19.8 GB: 45-of-55
Model Architecture	Qwen3_5MoeForCausalLM
License	mit
Context Length	262144
Model Max Length	262144
Transformers Version	5.15.0.dev0
Vocabulary Size	248320

Rank the Qwen3.8 2.4T A95B NVFP4 REAP 25 capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation