LLM EXPLORER 58,608 MODELS INDEXED

Molmo2 8B by allenai

By allenai · 230869 downloads
HF Repo ↗ Copy URL

Molmo2 8B is an open-source language model by allenai. Features: 8b LLM, VRAM: 34.5GB, License: apache-2.0, LLM Explorer Score: 0.46, ELO: 1329.

Base model:finetune:qwen/qwen3...   Base model:qwen/qwen3-8b   Conversational   Custom code Dataset:allenai/molmo2-askmode...   Dataset:allenai/molmo2-cap Dataset:allenai/molmo2-multiim... Dataset:allenai/molmo2-multiim... Dataset:allenai/molmo2-synmult... Dataset:allenai/molmo2-videoca... Dataset:allenai/molmo2-videopo... Dataset:allenai/molmo2-videosu... Dataset:allenai/molmo2-videotr...   En   Image-text-to-text   Molmo   Molmo2   Multimodal   Olmo   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/allenai/Molmo2-8B 

Molmo2 8B Benchmarks

LMSys ELO 1329 vs 1272 (so35) 4.5%
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

Molmo2 8B Parameters and Internals

LLM NameMolmo2 8B
Repository 🤗https://huggingface.co/allenai/Molmo2-8B 
Base Model(s)  Qwen3 8B   google/siglip-so400m-patch14-384   Qwen/Qwen3-8B   google/siglip-so400m-patch14-384
Model Size8b
Required VRAM34.5 GB
Updated2026-08-09
Maintainerallenai
Model Typemolmo2
Model Files  5.0 GB: 1-of-8   4.6 GB: 2-of-8   4.6 GB: 3-of-8   4.6 GB: 4-of-8   4.6 GB: 5-of-8   4.6 GB: 6-of-8   4.0 GB: 7-of-8   2.5 GB: 8-of-8
Supported Languagesen
Model ArchitectureMolmo2ForConditionalGeneration
Licenseapache-2.0
Model Max Length131072
Transformers Version4.57.1
Tokenizer ClassQwen2Tokenizer
Padding Token<|endoftext|>
Errorsreplace

Best Alternatives to Molmo2 8B

Best Alternatives
Context / RAM
Downloads
Likes
MolmoWeb 8B0K / 34.5 GB117770