Hermes Model & Ecosystem Tracker

Evaluation Resources

• 🔥 New Eval Repo: NousResearch released the eval-Hermes-4-70B-nonreasoning repository providing open access to evaluation...

Hermes Agent's Mixture of Agents (MoA) lets users route through reference models + aggregator for stronger outputs via simple preset selection.

-...

NousResearch published the eval-Hermes-4-70B-nonreasoning repository, delivering a dedicated evaluation setup and file assets for tracking the non-reasoning variant's performance.

Tailscale Funnel provides HTTPS access to your Hermes Agent without port forwarding or reverse proxies.

• Setup flow: Enable the built-in API server...

Hermes Agent differentiates itself as a persistent, self-improving agent that maintains memory and builds reusable skills across sessions, unlike...

Agent Feature Releases

• 🔥 /learn Command Added: Hermes Agent now includes a /learn command that ingests directories, docs, or workflows to...

Hermes Agent's /learn command lets the agent ingest directories, URLs, conversations, or notes to author reusable SKILL.md files without...

A free 90-minute tutorial delivers structured guidance for integrating Hermes with general agents, featuring official skill installation, a five-step...

The new Hermes Web UI dashboard introduces login lock commands for secure agent access.

• hermes-web-ui clear-login-locks removes persisted IP lock...

Nous Research has released the Hermes Agent Profile Builder with Blank Slate mode, enabling users to define agent identity, model, skills, and MCP for streamlined configuration.

Model Releases

• 🔥 New GGUF Quantization: mradermacher/Nous-Hermes-Llama2-70b-i1-GGUF is now available for download and local inference via...

A new 44-minute episode of Intelligent Machines features Jeffrey Quesnelle, CTO of Nous Research and creator of Hermes AI. The discussion explores the project directly from its architect.

PowerAgentBench-SS tests whether LLM agents can execute engineering workflows like grid inspection, unlike prior benchmarks that fall short on practical execution.

Quick 5-step guide to connect Cursor MCP:

• Get your MCP URL and API Key
• Open the Hermes config file
• Add the Composio Connect MCP server
• Restart your Hermes agent
• Done!

Hermes shifts from standalone agents to a self-evolving AI operating system, anchored by persistent memory.

• Self-Evolution Engine: Agents detect...

A quantized GGUF release of the older Nous-Hermes-Llama2-70b just dropped, enabling straightforward local inference through standard libraries and apps. This expands access for open-source users running Hermes models on consumer hardware.

Benchmark Comparisons

• 🔥 GLM 5.2 vs Hermes 4 70B: A direct benchmark comparison between GLM 5.2 and Hermes 4 70B is now available covering...

A fresh head-to-head analysis pits Hermes 4 70B against GLM 5.2 across benchmarks, pricing, context length, and key features to guide model selection.

Connect external tools like file systems, GitHub, and APIs to Hermes Agents via MCP Server with this practical tutorial.

• Covers MCP basics, server...

New video offers developers a detailed walkthrough of Hermes agent internals.

• Architecture overview: Covers the agent loop, context construction,...