What are the hardware requirements for Qwen3.5 122B A10B?

Qwen3.5 122B A10B requires approximately 248.6 GB of VRAM. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Qwen3.5 122B A10B and how large is it?

Qwen3.5 122B A10B is developed by SwinliQ-AIs, a model with 122b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Qwen3.5 122B A10B perform on standard benchmarks?

Qwen3.5 122B A10B has the following published scores: LMSYS ELO 1417. Compare against reference models on this page or on the LLM Explorer leaderboards.

Qwen3.5 122B A10B by SwinliQ-AIs — VRAM 248.6GB

Name: Qwen3.5 122B A10B
Author: SwinliQ-AIs

Qwen3.5 122B A10B is an open-source language model by SwinliQ-AIs. Features: 122b LLM, VRAM: 248.6GB, License: apache-2.0, LLM Explorer Score: 0.39, ELO: 1417.

Conversational Endpoints compatible Image-text-to-text Qwen3 5 moe Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/SwinliQ-AIs/Qwen3.5-122B-A10B

Qwen3.5 122B A10B Benchmarks

LMSys ELO 1417 vs 1272 (so35) 11.4%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Qwen3.5 122B A10B Parameters and Internals

LLM Name	Qwen3.5 122B A10B
Repository 🤗	https://huggingface.co/SwinliQ-AIs/Qwen3.5-122B-A10B
Model Size	122b
Required VRAM	248.6 GB
Updated	2026-08-09
Maintainer	SwinliQ-AIs
Model Type	qwen3_5_moe
Model Files	6.4 GB: 1-of-39 6.4 GB: 2-of-39 6.4 GB: 3-of-39 6.4 GB: 4-of-39 6.4 GB: 5-of-39 6.4 GB: 6-of-39 6.4 GB: 7-of-39 6.4 GB: 8-of-39 6.4 GB: 9-of-39 6.4 GB: 10-of-39 6.4 GB: 11-of-39 6.4 GB: 12-of-39 6.4 GB: 13-of-39 6.4 GB: 14-of-39 6.4 GB: 15-of-39 6.4 GB: 16-of-39 6.4 GB: 17-of-39 6.4 GB: 18-of-39 6.4 GB: 19-of-39 6.4 GB: 20-of-39 6.4 GB: 21-of-39 6.4 GB: 22-of-39 6.4 GB: 23-of-39 6.4 GB: 24-of-39 6.4 GB: 25-of-39 6.4 GB: 26-of-39 6.4 GB: 27-of-39 6.4 GB: 28-of-39 6.4 GB: 29-of-39 6.4 GB: 30-of-39 6.4 GB: 31-of-39 6.4 GB: 32-of-39 6.4 GB: 33-of-39 6.4 GB: 34-of-39 6.4 GB: 35-of-39 6.4 GB: 36-of-39 6.4 GB: 37-of-39 6.4 GB: 38-of-39 5.4 GB: 39-of-39
Model Architecture	Qwen3_5MoeForConditionalGeneration
License	apache-2.0
Model Max Length	262144
Transformers Version	4.57.0.dev0
Tokenizer Class	Qwen2Tokenizer
Padding Token	<\|endoftext\|>
Errors	replace

Rank the Qwen3.5 122B A10B 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

Best Alternatives to Qwen3.5 122B A10B

Best Alternatives	Context / RAM	Downloads	Likes
Qwen3.5 122B A10B NVFP4	0K / 83.5 GB	248732	49
Qwen3.5 122B A10B	0K / 248.6 GB	2139628	605
Qwen3.5 122B A10B	0K / 248.6 GB	1680	8
Qwen3.5 122B A10B Abliterated	0K / 248.6 GB	61	5
Qwen3.5 122B A10B FP8	0K / 126.3 GB	1271983	112
Qwen3.5 122B A10B NVFP4	0K / 74 GB	18267	18
...en3.5 122B A10B Int4 AutoRound	0K / 72 GB	142311	45
AREX Base	0K / 164.2 GB	395	21
T1 122B A10B	0K / 206.4 GB	41	4
...2.6 Destill Healed Abliterated	0K / 245.1 GB	1143	15

Note: green Score (e.g. "73.2") means that the model is better than SwinliQ-AIs/Qwen3.5-122B-A10B.