Lens Integrations

​Moondream is an open source vision language model designed for efficient image understanding across various devices, including servers, PCs, mobile phones, and edge devices. It offers two primary variants, Moondream 2B, a 1.9-billion-parameter model providing robust performance for general-purpose tasks, and Moondream 0.5B, a compact 500-million-parameter model optimized for resource-constrained hardware. Both models support quantization formats like fp16, int8, and int4, allowing for reduced memory usage without significant performance loss. Moondream's capabilities include generating detailed image captions, answering visual queries, performing object detection, and pinpointing specific items within images. Its design emphasizes versatility and accessibility, enabling deployment across a wide range of platforms. ​

Photon is Moondream’s official high-performance inference engine, designed to run vision-language models efficiently across cloud, desktop, and edge environments while delivering real-time performance for production AI systems. It is built as a custom inference layer tightly integrated with the Moondream model architecture, using optimized scheduling, native image processing, and purpose-built CUDA kernels to maximize speed and efficiency. This co-designed approach allows Photon to significantly reduce latency compared to traditional VLM setups, enabling responsive interactions on edge devices and real-time throughput on server-grade hardware. It supports deployment across a wide range of NVIDIA GPUs, from embedded systems like Jetson devices to high-end multi-GPU servers, making it adaptable for diverse operational needs. It includes production-ready features such as automatic batching, prefix caching, and memory-efficient attention mechanisms.