LLM EXPLORER 56,604 MODELS INDEXED

GTA1 32B by Salesforce

By Salesforce · 191 downloads
HF Repo ↗ Copy URL

GTA1 32B is an open-source language model by Salesforce. Features: 32b LLM, VRAM: 66.7GB, Context: 125K, License: mit, LLM Explorer Score: 0.21.

  Arxiv:2507.05791   Computer-use-agent   Conversational   En   Endpoints compatible   Eval-results   Grounding   Gui   Image-text-to-text   Os-agent   Qwen2 5 vl   Region:us   Safetensors   Sharded   Tensorflow   Vlm
Model Card on HF 🤗: https://huggingface.co/Salesforce/GTA1-32B 

GTA1 32B Parameters and Internals

LLM NameGTA1 32B
Repository 🤗https://huggingface.co/Salesforce/GTA1-32B 
Model Size32b
Required VRAM66.7 GB
Updated2026-08-08
MaintainerSalesforce
Model Typeqwen2_5_vl
Model Files  4.9 GB: 1-of-14   4.7 GB: 2-of-14   4.8 GB: 3-of-14   5.0 GB: 4-of-14   4.9 GB: 5-of-14   4.8 GB: 6-of-14   5.0 GB: 7-of-14   4.8 GB: 8-of-14   4.9 GB: 9-of-14   4.7 GB: 10-of-14   4.9 GB: 11-of-14   4.7 GB: 12-of-14   4.8 GB: 13-of-14   3.8 GB: 14-of-14   0.0 GB
Supported Languagesen
Model ArchitectureQwen2_5_VLForConditionalGeneration
Licensemit
Context Length128000
Model Max Length128000
Transformers Version4.49.0
Tokenizer ClassTikTokenV3
Padding Token[PAD]
Vocabulary Size152064
Torch Data Typebfloat16

Best Alternatives to GTA1 32B

Best Alternatives
Context / RAM
Downloads
Likes
ScaleCUA 32B125K / 67.2 GB6622
Qwen2.5 VL 32B Instruct Bf16125K / 67.3 GB3045
Lingshu 32B 4bit125K / 18.8 GB612
Lingshu 32B 8bit125K / 35.5 GB271
Qwen2.5 VL 32B Instruct 4bit125K / 19.8 GB4414
Qwen2.5 VL 32B Instruct 8bit125K / 36.2 GB1834
Qwen2.5 VL 32B Instruct 6bit125K / 28 GB771