LLM EXPLORER 65,695 MODELS INDEXED

Flora Captioner by nvidia

By nvidia · 1351 downloads
HF Repo ↗ Copy URL

Flora Captioner is an open-source language model by nvidia. Features: 27b LLM, VRAM: 109.4GB, License: other, LLM Explorer Score: 0.34.

  Arxiv:2503.14734 Base model:finetune:qwen/qwen3...   Base model:qwen/qwen3.5-27b   Conversational   Data-engine   Endpoints compatible   Image-text-to-text   Qwen3.5   Qwen3 5   Region:us   Safetensors   Sharded   Tensorflow   Video   Video-captioning
Model Card on HF 🤗: https://huggingface.co/nvidia/Flora-Captioner 

Flora Captioner Parameters and Internals

LLM NameFlora Captioner
Repository 🤗https://huggingface.co/nvidia/Flora-Captioner 
Base Model(s)  Qwen/Qwen3.5-27B   Qwen/Qwen3.5-27B
Model Size27b
Required VRAM109.4 GB
Updated2026-10-26
Maintainernvidia
Model Typeqwen3_5
Model Files  10.5 GB: 1-of-11   9.6 GB: 2-of-11   10.7 GB: 3-of-11   10.7 GB: 4-of-11   10.7 GB: 5-of-11   10.7 GB: 6-of-11   10.7 GB: 7-of-11   10.3 GB: 8-of-11   10.6 GB: 9-of-11   10.7 GB: 10-of-11   4.2 GB: 11-of-11
Model ArchitectureQwen3_5ForConditionalGeneration
Licenseother
Model Max Length262144
Transformers Version5.8.1
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Torch Data Typefloat32
Errorsreplace

Best Alternatives to Flora Captioner

Best Alternatives
Context / RAM
Downloads
Likes
...wopus3.8 27B Flash Mlx OQ4e 1M1024K / 16.9 GB2111
Agens Pilot1024K / 54.7 GB863
Qwen3.8 27B NPU2256K /  GB2524
JEV 27B VL0K / 55.6 GB1525286966
Clef0K / 54.7 GB220432237
Qwen3.6 27B0K / 55.4 GB67281382197
Qwen3.8 27B0K / 55.6 GB28753
Qwen3.8 27B0K / 55.6 GB272721
Qwen3.8 27B MLX0K / 16 GB141118
Qwen3.6 27B NVFP40K / 22 GB1887525419
Note: green Score (e.g. "73.2") means that the model is better than nvidia/Flora-Captioner.