Grug 35B A3B is an open-source language model by kai-os. Features: 34.7b LLM, VRAM: 69GB, Context: 256K, License: apache-2.0, MoE, Instruction-Based, LLM Explorer Score: 0.33.
A3b Base model:finetune:qwen/qwen3... Base model:qwen/qwen3.6-35b-a3... Compact-reasoning Conversational Dataset:cl-from-nothing/code r... Dataset:hotdogs/uka-glm-5.2 Dataset:hsh-intelligence/verif... Dataset:kd13/codedebug-instruc... Dataset:madarabr/cortex-adapti... Dataset:samcheng0/lumia-reason... Dataset:scale-or-reason/genera... Endpoints compatible Finetuned Instruct Model-index Moe Qlora Qwen Qwen3.6 Qwen3 5 moe text Reasoning Region:us Safetensors Sharded Tensorflow
Model Card on HF 🤗: https://huggingface.co/kai-os/Grug-35B-A3B
Grug 35B A3B Benchmarks
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)
Grug 35B A3B Parameters and Internals
| LLM Name | Grug 35B A3B |
| Repository 🤗 | https://huggingface.co/kai-os/Grug-35B-A3B |
| Base Model(s) | |
| Model Size | 34.7b |
| Required VRAM | 69 GB |
| Updated | 2026-07-03 |
| Maintainer | kai-os |
| Model Type | qwen3_5_moe_text |
| Instruction-Based | Yes |
| Model Files | |
| Model Architecture | Qwen3_5MoeForCausalLM |
| License | apache-2.0 |
| Context Length | 262144 |
| Model Max Length | 262144 |
| Transformers Version | 5.12.1 |
| Tokenizer Class | Qwen2Tokenizer |
| Padding Token | <|endoftext|> |
| Vocabulary Size | 248320 |
| Errors | replace |
Rank the Grug 35B A3B Capabilities
🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟
What open-source LLMs or SLMs are you in search of? 54752 in total.
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a