LLM EXPLORER 58,133 MODELS INDEXED

Zira Researcher by 0xvoid0000

By 0xvoid0000 · 11404 downloads
HF Repo ↗ Copy URL

Zira Researcher is an open-source language model by 0xvoid0000. Features: 4b LLM, VRAM: 8.4GB, Context: 256K, License: apache-2.0, LLM Explorer Score: 0.29.

  0xvoid Base model:finetune:qwen/qwen3...   Base model:qwen/qwen3.5-4b   Conversational   En   Finetuned   Long-context   Qwen3 5   Qwen3 5 text   Reasoning   Region:us   Research   Safetensors   Self-correction   Trl   Zira
Model Card on HF 🤗: https://huggingface.co/0xvoid0000/zira-researcher 

Zira Researcher Parameters and Internals

LLM NameZira Researcher
Repository 🤗https://huggingface.co/0xvoid0000/zira-researcher 
Base Model(s)  Qwen/Qwen3.5-4B   Qwen/Qwen3.5-4B
Model Size4b
Required VRAM8.4 GB
Updated2026-08-11
Maintainer0xvoid0000
Model Typeqwen3_5_text
Model Files  8.4 GB
Supported Languagesen
Model ArchitectureQwen3_5ForCausalLM
Licenseapache-2.0
Context Length262144
Model Max Length262144
Transformers Version5.2.0
Tokenizer ClassTokenizersBackend
Padding Token<|im_end|>
Vocabulary Size248320
Errorsreplace

Best Alternatives to Zira Researcher

Best Alternatives
Context / RAM
Downloads
Likes
Agents A1 4B Fable Preview256K / 8.4 GB1030
...ts A1 4B Fable Preview Heretic256K / 8.4 GB22852
Qwen35 4B T13 Opd Rkl Ckpt50256K / 8.4 GB4670
...nts A1 4B Kimi Preview Heretic256K / 8.4 GB2870
Qwen3.5 4B EU256K / 9.7 GB870
Qwen3.5 4B INT8256K / 4.8 GB2052
Qwen35 4B Soyuz Merged256K / 8.4 GB733
Retrico Lm 4B Finetuned V4256K / 0 GB1430
Holo 3.1 4B Coder256K / 16.8 GB411
Qwen3.5 4B INT8256K / 4.8 GB1112
Note: green Score (e.g. "73.2") means that the model is better than 0xvoid0000/zira-researcher.