Show HN: ilovellm – Private, in-browser LLM tools (ilovellm.com)

ilovellm is a new collection of free, in-browser LLM and multimodal tools that run entirely on-device, announced on Hacker News. The suite packs speech and text capabilities (Whisper-based audio-to-text, speech recognition with word-level timestamps and speaker segmentation, translation, language ID, sentiment/text classification, and sentence embeddings), lightweight TTS (promoted as state-of-the-art under 25MB), and a broad set of vision and interaction features (image classification, object detection, segmentation including click-to-segment, hand-gesture recognition, and real-time facial landmark tracking via webcam). For the AI/ML community this matters because it lowers barriers to private, low-latency, offline AI—no server round-trips or data uploads are required, which preserves user privacy and reduces infra costs. Running embeddings, classification and speech/vision pipelines client-side also enables rapid prototyping, edge deployments, and demos that are more accessible to end users. The trade-offs are the usual browser/edge constraints: limited compute, model size/accuracy trade-offs, and reliance on client hardware/Web APIs. Still, the package is significant as a practical, privacy-first alternative for many applications where cloud models are unnecessary or undesirable.