Hermes Agent v0.8.0 ships notify-on-complete, live model switching, and free Xiaomi MiMo
TL;DR
Hermes Agent v0.8.0 landed April 8, 2026. The Intelligence Release adds background task auto-notifications so the agent no longer blocks on long-running work, live /model switching across CLI and every messaging channel, free Xiaomi MiMo v2 Pro on Nous Portal, native Google AI Studio provider, and MCP OAuth 2.1 with OSV malware scanning.
What changed
What shipped
Hermes Agent v0.8.0 (The Intelligence Release) dropped April 8, 2026. Five changes of note:
Background task auto-notifications
A new notify_on_complete capability lets long-running tasks (training runs, test suites, deploys) notify the agent when they finish, instead of blocking the agent on polling. The agent continues with other work and picks up the result when it lands. For cron-agent workloads this is a durability unlock.
Free Xiaomi MiMo v2 Pro
Nous Portal subscribers get MiMo v2 Pro free for auxiliary tasks: compression, vision, summarization. Pricing and free-tier gating are displayed right in the model picker so you know what you are spending before you spend it.
Live model switching
The /model command now works mid-session. Switch models across CLI, Telegram, Discord, Slack, and any gateway without starting a new conversation. Interactive inline-button model pickers on Telegram and Discord make the switch feel native rather than forced through slash syntax.
Native Google AI Studio provider
Gemini is now a first-class provider, not a proxy through OpenRouter. Self-optimised GPT and Codex tool-use guidance landed in the same release, patching 5 failure modes that the team identified through automated behavioural benchmarking.
Extension + timeout hardening
- •MCP OAuth 2.1 PKCE for secure third-party auth flows.
- •OSV malware scanning of MCP extension packages before install.
- •Inactivity-based agent timeouts idle-only, not wall-clock, so long-running tasks do not get killed just because the clock is running.
- •Approval buttons on Slack and Telegram for destructive tool calls, replacing text-based "type approve" dances.
Availability
Standard Hermes upgrade path. No breaking config changes; existing MCP setups continue to work (OSV scanning runs on next install, not existing packages).
Who this matters for
- Vibe Builder: Free MiMo for compression and vision tasks on Nous Portal. Live /model switching mid-conversation. Inline-button model pickers on Telegram and Discord.
- Basic User: Approval buttons on Slack and Telegram replace 'type approve' prompts. Inactivity-based timeouts keep long tasks alive rather than killing them on wall-clock.
- Developer: notify_on_complete is the durability win for cron-agent workloads. MCP OAuth 2.1 PKCE and OSV malware scanning bring MCP security posture up to where it should be.
What to watch next
v0.8.0's notify_on_complete is the feature that matters most for real production use. Every long-running agent workflow I have seen hits the same wall: the agent either blocks uselessly while a 20-minute task runs, or uses a polling loop that burns context and credits. Switching to event-driven notifications changes what is practical for cron-agent workloads.
Free MiMo v2 Pro on Nous Portal is the economics story. Most "self-improving" agents end up burning money on auxiliary model calls for housekeeping (summarisation, compression, vision). Including a free high-quality auxiliary model with a subscription moves Hermes from expensive to defensibly cheap for anyone running it seriously.
Live model switching mid-session removes one of the most common papercuts: having to restart a conversation to try a different model on the same task. /model on Telegram with inline buttons is the small UX move that makes the agent feel like a modern product rather than a developer tool.
The MCP OAuth 2.1 plus OSV scanning combination is the MCP security posture catching up to where it needed to be. MCP's supply-chain attack surface is real; OSV package scanning before install is table stakes for any framework that encourages MCP extensions. Nice to see Hermes get there before an incident forces the conversation.
by Harsh Desai
About hermes-agent
View the full hermes-agent page →All hermes-agent updatesMore from hermes-agent
- FeatureHermes Agent v0.4.0 adds OpenAI-compatible API, 6 messaging platforms, and @file context
Hermes Agent v0.4.0 (The Platform Expansion Release) shipped March 24, 2026. It exposes Hermes as a drop-in OpenAI-compatible /v1/chat/completions endpoint, adds 6 new messaging adapters (Signal, DingTalk, SMS via Twilio, Mattermost, Matrix, Webhook) bringing the count to 9, introduces @file and @url context injection, adds 4 new inference providers, and enables gateway prompt caching by default.
- FeatureHermes Agent v0.5.0 expands to 400+ models and adds supply-chain hardening
Hermes Agent v0.5.0 (The Hardening Release) shipped March 28, 2026. Nous Portal now routes to 400+ models through one endpoint. Hugging Face becomes a first-class inference provider. Telegram gets private chat topics with per-topic skill binding. Plugin lifecycle hooks activate. Supply-chain hardening removes a compromised dep, pins version ranges, and adds CI scanning.
- FeatureHermes Agent v0.6.0 lands profiles, MCP server mode, and official Docker container
Hermes Agent v0.6.0 (The Multi-Instance Release) shipped March 30, 2026. Profiles let you run multiple isolated Hermes instances from one install. MCP Server Mode exposes the agent's conversations and sessions to Claude Desktop, Cursor, VS Code, and any MCP client. Official Docker container, two new messaging platforms (Feishu/Lark, WeCom), and an ordered fallback provider chain round out the release.