Moondream: Moondream 2

SmolVLM2 is a compact multimodal vision-language model developed by the Hugging Face TB Research team, released in February 2025 under the Apache 2.0 license. It is designed for efficient image and video understanding on resource-constrained hardware, with model variants ranging from 256M to 2.2B parameters. SmolVLM2 processes images, multi-image inputs, and video alongside text queries to generate text outputs for tasks including visual question answering, image captioning, and OCR.SmolVLM2 is designed for on-device and edge deployment, requiring substantially less GPU memory than comparable multimodal models. It supports standard fine-tuning pipelines via the Hugging Face transformers library and quantization through bitsandbytes. SmolVLM2 is suited for applications where a capable vision-language model is needed without full server-scale infrastructure.

PaliGemma 2 is a vision-language model released in December 2024 by Google DeepMind. It pairs the SigLIP-So400m vision encoder with the Gemma 2 language model family, extending the original PaliGemma architecture with stronger language capabilities and a wider set of transfer benchmarks. The model is designed primarily as a fine-tuning base rather than a chat-optimized assistant. Google releases pretrained "PT" checkpoints intended for task-specific adaptation rather than direct out-of-the-box use.PaliGemma 2 accepts an image paired with a text prompt and generates natural language output, supporting image captioning, visual question answering, optical character recognition, document understanding, object detection and segmentation (with appropriate fine-tuning), and a range of specialized vision-language tasks. The model is released at three parameter sizes (3B, 10B, and 28B), built on the Gemma 2 2B, 9B, and 27B language backbones. Each size is available at three input resolutions: 224, 448, and 896 pixels. Alongside the base PT checkpoints, Google released PaliGemma 2 Mix variants that have been tuned on a mixture of downstream tasks to provide stronger out-of-the-box performance for common applications such as OCR and document parsing. PaliGemma 2 is distributed under the Gemma license, a custom license from Google that permits commercial use subject to the terms of the Gemma Prohibited Use Policy.

PaliGemma is a vision-language model released in May 2024 by Google, built by pairing the SigLIP-So400m vision encoder with the Gemma 2B language model. It is designed primarily as a compact, transfer-friendly base model for fine-tuning to downstream vision-language tasks, rather than as a chat-optimized assistant. PaliGemma draws architectural inspiration from the PaLI-3 model at Google Research, applying a similar encoder-decoder approach at a smaller and more accessible parameter scale.PaliGemma accepts an image together with a text prompt and generates text output, supporting image captioning, visual question answering, optical character recognition, object detection, referring expression segmentation, and a range of related vision-language tasks when fine-tuned on task-specific data. The model is released at three input resolutions (224, 448, and 896 pixels), with higher resolutions providing stronger performance on tasks requiring fine visual detail such as OCR and document understanding. Google released pretrained (PT) checkpoints intended as fine-tuning bases, along with Mix variants that have been fine-tuned on a mixture of downstream tasks for direct use without additional training. PaliGemma is distributed under the Gemma license, a custom license from Google that permits commercial use subject to the terms of the Gemma Prohibited Use Policy. It was succeeded by PaliGemma 2 in December 2024, which extends the architecture to larger Gemma 2 language backbones at 3B, 10B, and 28B parameter sizes.

LLaVA-1.5 is an open-source large multimodal model released in October 2023 by researchers at the University of Wisconsin-Madison and Microsoft Research. It builds on the original LLaVA architecture by introducing targeted refinements: switching the vision encoder to CLIP-ViT-L at 336-pixel resolution, replacing the projection layer with a two-layer MLP, and adding academic-task-oriented visual question answering data with response formatting prompts during training. These modifications achieve state-of-the-art performance across 11 benchmarks at release, with training completing in approximately one day on a single 8-A100 node.The model accepts an image paired with a text prompt and generates natural language responses, supporting visual question answering, image captioning, and open-ended visual conversation. LLaVA-1.5 is available in 7B and 13B parameter variants built on the Vicuna language model, and is distributed under the Llama 2 Community License due to its Llama-2-based foundation. The original LLaVA paper was presented as an oral at NeurIPS 2023. Subsequent releases in the series (LLaVA-NeXT (LLaVA-1.6), LLaVA-NeXT-Video, and LLaVA-OneVision) are separate models with their own release pages and build on this foundation with expanded OCR, video, and multi-image capabilities.

Florence-2, introduced by Microsoft Research at CVPR 2024, is an open-source vision-language foundation model designed to unify diverse computer vision tasks within a single sequence-to-sequence framework. Unlike traditional models that specialize in specific tasks, Florence-2 accepts both images and text prompts and outputs text for tasks such as captioning, object detection, segmentation, OCR, and region-based grounding. It comes in two sizes—Florence-2-base (~230M parameters) and Florence-2-large (~770M parameters)—and is trained on FLD-5B, a large dataset of ~126M images with ~5.4B annotations.The model demonstrates strong zero-shot and fine-tuned performance, often rivaling larger vision-language systems while remaining lightweight and efficient. Released under the MIT license, all weights are publicly available, making it accessible for fine-tuning and deployment in applications like VQA, content tagging, accessibility, and research. Florence-2’s compact design, versatility, and openness position it as a practical alternative to larger proprietary multimodal models.