AI tools for Vision

Atoms.dev is a multi‑agent product development platform that transforms plain‑language ideas into fully functional, production‑ready web and mobile applications, without requiri

Lumi represents a fundamentally new approach to web development, shifting the paradigm from traditional coding to intuitive conversational creation. Instead of wrestling with co

AirMusic is an innovative platform leveraging advanced algorithms to allow users to generate completely original music in a matter of seconds. By simply providing input regardin

Readdy is an innovative design-to-code platform that bridges the gap between creative vision and technical implementation by transforming natural language descriptions into prof

Xiaomi-Robotics-1 is a vision-language-action foundation model for mobile manipulation. Xiaomi trained the model on more than 100,000 hours of real-world manipulation trajectori

ReLift reconstructs dense 3D point trajectories from a monocular video. It combines the output of a 2D tracker with a video geometry estimator to recover how points move in spac

Cactus Mirage is a compact vision model for local image understanding and structured extraction. It can describe a scene, locate objects, read text and produce masks or structur

LFM3 is Liquid AI's family of compact models for specialized and on-device intelligence. Its first release includes text, vision, audio, omni, and decision variants built around

SceneBloom creates realistic indoor environments for training embodied agents. It translates room descriptions into varied furniture layouts and simulation-ready scenes, coverin

PhysAg3D turns an object image into a detailed 3D asset with object-specific interactions. It combines generated appearance and geometry with reasoning about parts, joints, and

WeVisOne-VL is a vision-language model that handles fine-grained visual perception from individual objects down to pixels. A Qwen3-VL backbone predicts structured coordinate tok

Iris-3B is a pixel-space generative model designed to serve as both an image generator and a general visual learner. It works directly in pixel space rather than compressing ima

PixelUMM is a unified model for understanding and generating images and video directly in pixel space. A single decoder-only transformer processes image patches and video patche

ORBIT is a benchmark for evaluating camera-pose estimation in difficult real-world video. It uses panoramic footage to create perspective-view test clips with recoverable refere

Show-Harness lets vision-language models control robots through a compact set of semantic action symbols. It connects high-level visual reasoning with fine-grained physical deci

Marigold V2 converts a single image into detailed depth and other dense visual predictions. Its model family covers ordinary depth, geometry behind transparent surfaces, surface

Fibo 1.5 is Bria JSON-native image model for controlled visual generation. It describes composition through structured attributes such as lighting, camera, color, and scene cont

Lucida reconstructs indoor captures as complete, editable object assets positioned in the observed scene. It separates the work into parsing, generation, and placement, aiming t

Nex-N2.5 is a model family for sustained agent work across computers, browsers, coding tasks, and knowledge workflows. Its Mini and Pro models combine vision with action-oriente

Qwen-Drive-1.0-4B combines driving-scene understanding, 3D perception, and motion planning in a unified vision-language framework. It can answer questions about a scene, expose

DeepSeek-V4-Flash-Vision-Exp is an experimental multimodal model that adds visual understanding to the DeepSeek-V4-Flash family. It accepts images and text for reasoning and age

Qwen3.8-27B is a dense 27-billion-parameter vision-language model for coding, professional work, research, and long-horizon agentic tasks. It understands both text and visual in

GenCeption is a research model that repurposes a pretrained video generation model into a unified feed-forward vision learner. Instead of training a separate model head for each

Inkling is Thinking Machines Lab's open-weights multimodal foundation model for customization, fine-tuning, and agentic workflows. The release describes a Mixture-of-Experts tra