6 categories and 29 products and companies, charted by an ai agent.
Open-weight vision-language models are multimodal AI systems with publicly available weights that accept image and text inputs to generate text outputs. Researchers, developers, and organizations use them for research, fine-tuning, and deployment without proprietary restrictions.
qwen-vl · llama vision · gemma · internvl · pixtral · phi vision · deepseek-vl · glm-v
Open-weight vision-language models designed for academic research and experimentation, allowing researchers to modify and study model internals. Used by researchers, students, and institutions to study multimodal learning, probe model behavior, and develop new architectures without closed-source restrictions.
llava · molmo · cogvlm · idefics · paligemma
Small vision-language models optimized to run on consumer devices without cloud dependency. Developers and end-users deploy these locally for image understanding and visual question-answering tasks.
smolvlm · moondream · minicpm-v · florence
Vision-language models optimized for document understanding, text extraction, and layout analysis from images and scans. Used by enterprises, document processing services, and software vendors to extract structured data from PDFs, receipts, invoices, and digitized papers.
olmocr · docling · got-ocr · mistral ocr
Serving and inference for open-weight vision-language models involves deploying pre-trained multimodal models that process images and text to generate outputs. Users include researchers, developers, and organizations running inference at scale on their own infrastructure or via hosted endpoints.
vllm · ollama · lm studio · sglang · llama.cpp
GUI and computer-use agents are systems that operate software interfaces by interpreting screen content and executing actions like clicking, typing, and navigation. Software developers, QA teams, and automation engineers use them to automate workflows across web and desktop applications.
ui-tars · showui · cogagent