LLM EXPLORER 57,252 MODELS INDEXED

P1 VL 30B A3B by PRIME-RL

By PRIME-RL · 27 downloads
HF Repo ↗ Copy URL

P1 VL 30B A3B is an open-source language model by PRIME-RL. Features: 30b LLM, VRAM: 62.3GB, License: apache-2.0, LLM Explorer Score: 0.22.

  Arxiv:2602.09443   Conversational   Endpoints compatible   Image-text-to-text   Olympiad   Physics   Qwen3 vl moe   Reasoning   Region:us   Safetensors   Science   Sharded   Tensorflow   Vision   Vision-language
Model Card on HF 🤗: https://huggingface.co/PRIME-RL/P1-VL-30B-A3B 

P1 VL 30B A3B Parameters and Internals

LLM NameP1 VL 30B A3B
Repository 🤗https://huggingface.co/PRIME-RL/P1-VL-30B-A3B 
Model Size30b
Required VRAM62.3 GB
Updated2026-08-08
MaintainerPRIME-RL
Model Typeqwen3_vl_moe
Model Files  4.4 GB: 1-of-13   5.0 GB: 2-of-13   5.0 GB: 3-of-13   5.0 GB: 4-of-13   5.0 GB: 5-of-13   5.0 GB: 6-of-13   5.0 GB: 7-of-13   5.0 GB: 8-of-13   5.0 GB: 9-of-13   5.0 GB: 10-of-13   5.0 GB: 11-of-13   5.0 GB: 12-of-13   2.9 GB: 13-of-13
Model ArchitectureQwen3VLMoeForConditionalGeneration
Licenseapache-2.0
Model Max Length262144
Transformers Version4.57.0.dev0
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to P1 VL 30B A3B

Best Alternatives
Context / RAM
Downloads
Likes
Qwen3 VL 30B A3B Instruct0K / 62.3 GB488106591
Qwen3 VL 30B A3B Instruct FP80K / 32.2 GB734631115
Bu 30B A3b Preview0K / 62.3 GB1035263
UI Venus 1.5 30B A3B0K / 62.3 GB110735
Qwen3 VL 30B A3B Instruct0K / 62.3 GB5153
Gelato 30B A3B0K / 62.1 GB3034
Qwen3 VL 30B A3B Instruct Bf160K / 62.3 GB874
Qwen3 VL 30B A3B Thinking Bf160K / 62.3 GB421
Qwen3 VL 30B A3B Instruct 4bit0K / 18.3 GB9258
Bu 30B A3b Preview 4bit0K / 18.3 GB524
Note: green Score (e.g. "73.2") means that the model is better than PRIME-RL/P1-VL-30B-A3B.