Hermes v2026.4.23 Update: Hermes Agent v0.11.0 (2026.4.23)

This release is all about making Hermes Agent easier to use, more flexible, and smarter — with a brand-new interface, native support for more AI providers, and a whole new messaging platform.

What’s New

A Fresh Look and Feel
The interactive command-line tool (hermes --tui) has been completely rebuilt using modern web technology (React/Ink). Think of it like upgrading from a basic text editor to a full-featured IDE. You get a sticky input box that stays at the bottom, live streaming of responses with clipboard support, and a status bar showing how long each turn took and which git branch you’re on. There’s even a light theme for daytime use and a way to see what sub-agents are doing in real-time.

Pluggable AI Transports
Previously, all AI provider connections were hardcoded into one file. Now, they’re split into separate, interchangeable modules — like having universal adapters for different power outlets. This makes it much easier to add new providers and keep existing ones stable.

Native AWS Bedrock Support
You can now use Amazon’s Bedrock service directly, without any extra configuration layers. It’s like having a direct highway instead of a bumpy back road.

Five New Inference Paths
Hermes now works with NVIDIA NIM, Arcee AI, Step Plan, Google Gemini (via CLI OAuth), and Vercel’s AI gateway. Plus, Gemini traffic now goes through Google’s native AI Studio API for better speed. That’s like adding five new fuel types to your car — you can pick whatever’s cheapest or fastest at the moment.

GPT-5.5 via Codex OAuth
OpenAI’s latest reasoning model is now available through your existing ChatGPT Codex login. Even better, the model picker automatically discovers new OpenAI releases — no waiting for a manual update.

QQBot — 17th Messaging Platform
You can now connect Hermes to QQ (China’s popular messaging app) using the official API v2, with a simple QR code scan to set it up. That brings the total supported platforms to 17, covering everything from Slack to Discord to WhatsApp.

Improvements

Transport layer refactoring — cleaner code, fewer bugs, easier to maintain
Better Gemini performance — faster responses through native API routing
Live model discovery — new AI models appear in your picker automatically

Fixes (Highlights)

Deferred features from v0.10.0 are now all included (except the Nous Tool Gateway, which shipped earlier)
Various stability fixes across the new transport layer and TUI

What This Means for You

If you’re a daily user, the new TUI will feel like a breath of fresh air — smoother, more informative, and less cluttered. If you’re a developer, the pluggable transport architecture makes it trivial to add your own provider. And if you’re a power user, the five new inference paths plus GPT-5.5 give you more choices than ever before.

This release is a big step toward making Hermes Agent the most flexible, user-friendly open-source AI agent framework out there. Whether you’re chatting via QQ, coding with AWS Bedrock, or just enjoying the new interface — there’s something here for everyone.

📎 Official release notes: https://github.com/NousResearch/hermes-agent/releases