hermes-ai.net
Read docs →Announcements, milestones, and updates from the Hermes Agent project.
Hermes now ships a native desktop app for macOS, Linux, and Windows with one-click install, in-app self-update, drag-and-drop chat files, an inline model picker, concurrent profiles, Simplified Chinese UI, remote gateway sign-in, a fuller web admin panel, Quick Setup through Nous Portal, NVIDIA/skills as a trusted tap, fuzzy model search, and /undo.
Rather than repeating every Desktop launch article, this selected external report captures the broader press angle: Hermes Desktop is available as a public preview on macOS, Windows, and Linux, with persistent memory, natural-language task planning, multi-channel operation, web search, image generation, speech, and sandboxed execution backends.
The v0.15.1 patch fixes the loopback dashboard 401 reload storm, makes Docker dashboard insecure mode an explicit opt-in, restores Skills page source/category pills, and improves MCP bare-command resolution inside Docker containers.
The Velocity Release refactors run_agent.py into focused agent modules, turns Kanban into a multi-agent platform, cuts cold-start and per-turn overhead, rebuilds session_search to be fast and free, adds promptware defenses, skill bundles, new image providers, a Nous-approved MCP catalog, OpenHands orchestration, ntfy messaging, and deep xAI support.
Hermes adds SuperGrok OAuth with a 1M-context Grok path, an OpenAI-compatible local proxy for OAuth-backed providers, first-class X search, end-to-end Microsoft Teams support, lighter installs, PyPI packaging, faster browser CDP calls, LINE and SimpleX Chat, Windows beta support, new optional skills, and Hugging Face as a trusted skills tap.
The Tenacity Release adds durable multi-agent Kanban, persistent /goal tracking, Checkpoints v2, gateway auto-resume, no_agent cron watchdog mode, stronger security defaults, Google Chat as another platform, provider plugins, and seven new i18n locales.
The Hermes release ships an autonomous background Curator that silently keeps your skill library sharp, a substantially upgraded self-improvement loop, four new inference providers, native Microsoft Teams integration, and a 57% reduction in TUI cold start time.
The Hermes Agent repository crossed 100,000 GitHub stars, making it one of the fastest-growing open-source AI agent projects. Thank you to every contributor, user, and star-giver who made this milestone possible.
v2026.4.16 introduces the Tool Gateway, enabling web search, AI image generation, text-to-speech, and browser automation for Nous Portal subscribers. Hermes can now see, speak, and browse the web on your behalf.
v2026.4.13 adds native support for Android via Termux, extends messaging to iMessage and WeChat, and introduces Fast Mode for OpenAI and Anthropic providers to cut response latency significantly.
Hermes users can now access over 400 models through Nous Portal, including the full Hugging Face model catalog. This release also adds Telegram Private Chat Topics and Nix flake support for reproducible, declarative Hermes installations.
MarkTechPost covered Hermes Agent as an open-source autonomous system built to address memory decay and environmental isolation, highlighting skill documents, persistent workspace access, and local, Docker, SSH, Singularity, and Modal execution backends.
