LLM EXPLORER 60,384 MODELS INDEXED

Qwen3.6 35B A3B Heretic REAP 0.30 by groxaxo

By groxaxo · 66 downloads
HF Repo ↗ Copy URL

Qwen3.6 35B A3B Heretic REAP 0.30 is an open-source language model by groxaxo. Features: 25.1b LLM, VRAM: 50.2GB, Context: 256K, License: apache-2.0, MoE, LLM Explorer Score: 0.25.

  Arxiv:2510.13999   Conversational   Endpoints compatible   Expert-pruning   Linear-attention   Mixture-of-experts   Moe   Pruning   Qwen3   Qwen3 5 moe text   Reap   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/groxaxo/Qwen3.6-35B-A3B-Heretic-REAP-0.30 

Qwen3.6 35B A3B Heretic REAP 0.30 Parameters and Internals

LLM NameQwen3.6 35B A3B Heretic REAP 0.30
Repository 🤗https://huggingface.co/groxaxo/Qwen3.6-35B-A3B-Heretic-REAP-0.30 
Base Model(s)  Qwen/Qwen3.5   Qwen/Qwen3.5
Model Size25.1b
Required VRAM50.2 GB
Updated2026-07-28
Maintainergroxaxo
Model Typeqwen3_5_moe_text
Model Files  49.4 GB: 1-of-2   0.8 GB: 2-of-2   1.2 GB
Model ArchitectureQwen3_5MoeForCausalLM
Licenseapache-2.0
Context Length262144
Model Max Length262144
Transformers Version5.12.1
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Vocabulary Size248320
Errorsreplace

Best Alternatives to Qwen3.6 35B A3B Heretic REAP 0.30

Best Alternatives
Context / RAM
Downloads
Likes
Qwen3.6 Whittle 25B A3B256K / 49.9 GB6111
Qwen3.5 24.5B Reapped V1256K / 50.1 GB1183
Note: green Score (e.g. "73.2") means that the model is better than groxaxo/Qwen3.6-35B-A3B-Heretic-REAP-0.30.