LLM EXPLORER 60,519 MODELS INDEXED

Iris Mini by AllSpark-Research

By AllSpark-Research · 182 downloads
HF Repo ↗ Copy URL

Iris Mini is an open-source language model by AllSpark-Research. Features: 35b LLM, VRAM: 72.1GB, License: apache-2.0, LLM Explorer Score: 0.3.

  Agentic Base model:finetune:qwen/qwen3... Base model:qwen/qwen3.6-35b-a3...   Conversational   Deep-research   Endpoints compatible   Image-text-to-text   Moe   Qwen3.6   Qwen3 5 moe   Region:us   Safetensors   Search-agent   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/AllSpark-Research/Iris-mini 

Iris Mini Parameters and Internals

LLM NameIris Mini
Repository 🤗https://huggingface.co/AllSpark-Research/Iris-mini 
Base Model(s)  Qwen/Qwen3.6-35B-A3B   Qwen/Qwen3.6-35B-A3B
Model Size35b
Required VRAM72.1 GB
Updated2026-09-05
MaintainerAllSpark-Research
Model Typeqwen3_5_moe
Model Files  4.7 GB: 1-of-16   4.0 GB: 2-of-16   5.0 GB: 3-of-16   4.5 GB: 4-of-16   4.0 GB: 5-of-16   5.0 GB: 6-of-16   4.5 GB: 7-of-16   4.0 GB: 8-of-16   5.0 GB: 9-of-16   4.5 GB: 10-of-16   4.0 GB: 11-of-16   5.0 GB: 12-of-16   4.5 GB: 13-of-16   4.0 GB: 14-of-16   5.0 GB: 15-of-16   4.4 GB: 16-of-16
Model ArchitectureQwen3_5MoeForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version5.12.1
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Iris Mini

Best Alternatives
Context / RAM
Downloads
Likes
Ornith 1.5 35B A3B NPU2256K /  GB1531
Qwen3.6 35B A3B0K / 72.2 GB55408552648
Qwen3.5 35B A3B0K / 72.3 GB21585111482
Ornith 1.0 35B0K / 70.4 GB2983134506
Qwen3.6 35B A3B NVFP40K / 23.4 GB8788684482
KAT Coder V2.5 Dev0K / 69.7 GB17885537
Qwen3.6 35B A3B FP80K / 0.8 GB8984811343
Qwen AgentWorld 35B A3B0K / 69.2 GB66155665
Ornith 1.5 35B A3B0K / 72.1 GB1713201
Qwen3.6 35B A3B NVFP4 Fast0K / 23.6 GB46366198
Note: green Score (e.g. "73.2") means that the model is better than AllSpark-Research/Iris-mini.