Brave Leo

Only about 10 of the 25 top comments are actually about Leo; the rest are HackerNews threads about adjacent topics (Chrome Manifest V3 ad-blocking, running Mixtral/local LLMs on consumer GPUs and Nvidia Jetson boards, Apple Silicon memory bandwidth), which suggests the discussion energy around Leo is really about local-model feasibility, not the product itself. The genuine Leo users are mildly positive: 'Brave AI is simple and strong, as it reads websites content instead of using prestored data', 'a good brainstorming assistant', and one user runs it for real work — 'I using it to help configure my Nixos and code... I trust it's privacy and also doesn't make me too lazy.' Reliability is unproven: one user reports asking Leo its name and it said no; another asks how to remove the 'Search the web with AI' dialog box (UI annoyance); a Russian-language commenter argues privacy-first search engines return worse results because they lack behavioral data. On the brand's core promise, HN commenters split: one says 'a pinky promise from Brave or Mullvad is good enough' for casual privacy, while a reply pushes back that users lack options and expertise to verify such promises. Net: small, self-selected sample, no heavy usage reports, no premium-tier testimony.

No aggregate ratings were found for this product during the last harvest.