Video: "GLM 5.2 + Hermes AI SEO: 0 to 200 Clicks a Day!" by Julian Goldie on YouTube.

What GLM 5.2 actually is

GLM 5.2 is Z.ai's flagship open-weights model, released 13 June 2026 and available on Hugging Face under an MIT licence. The architecture is a Mixture of Experts with 744 billion total parameters and roughly 40 billion active per token — meaning the inference cost is much lower than the headline parameter count suggests. The context window is one million tokens, which is sufficient to hold a full website's worth of content, a large keyword database, or an extended agent conversation without truncating.

On the Artificial Analysis Intelligence Index v4.1, GLM 5.2 scored 51 — the highest of any open-weights model at the time of Julian Goldie's video. On long-horizon coding benchmarks it matches Claude Opus 4.8 and outscores GPT-5.5 in several structured task categories. The practical implication is that you can run frontier-level reasoning locally, on your own hardware, at no ongoing API cost.

The Infinite Video Engine

Julian Goldie's Infinite Video Engine is a Hermes Agent skill powered by GLM 5.2 that takes a single topic and produces a finished video without further input. The pipeline runs on Mac and covers script generation, scene composition, B-roll selection, caption overlay, and final render. The output is a complete short-form video ready for YouTube, Shorts, or social distribution — generated from one sentence.

The point is not to compete with polished branded video production. It is to collapse the time between "I have a topic" and "I have a published video asset" to the point where publishing daily video content becomes operationally feasible for a small team. Once the skill is configured, the only input required per video is the topic.

The Everywhere Engine

The Everywhere Engine is a companion Hermes skill that takes the same content and publishes it in the format expected by each search surface simultaneously. Running on GLM-5.2, it produces traditional web content optimised for Google's ranking signals alongside conversational versions structured to appear in ChatGPT browsing results and Perplexity AI answers. Rather than maintaining separate workflows for each platform, you configure the skill once and the agent handles the format differences.

This matters because AI Overviews in Google, ChatGPT with web browsing, and Perplexity all pull from different content signals. A single well-written page does not automatically rank in all three. The Everywhere Engine addresses that by generating platform-appropriate variants from a single source brief — the same research, adapted output.

What the traffic result shows

Julian Goldie demonstrated the system on a fresh domain. Starting from zero organic traffic and zero backlinks, the combination of Infinite Video Engine content and Everywhere Engine distribution reached 200 organic clicks per day. The demonstration used GLM 5.2 locally rather than any paid frontier API — the only cost was compute time on the host machine.

Two hundred clicks per day is not a large number for an established site, but the point of the demonstration is the economics: the same result previously required either significant editorial resource or paid API spend. GLM 5.2's quality level means the open-source option is now competitive on output quality, not just on price.

What this means for UK businesses

The release of a frontier-quality open-weights model with a permissive licence changes the cost structure of AI-assisted SEO. Teams that previously needed to budget for Anthropic or OpenAI API calls to run a coherent content workflow now have a credible local alternative. GLM 5.2 running inside Hermes Agent can handle keyword research, content generation, quality review, and multi-platform publishing from a single machine — the same tasks that previously required cloud API spend at every step.

The Hermes integration also means the model benefits from Hermes's persistent memory and tool-use layer. It is not just a language model you query for individual outputs; it is a component in a running agent loop that accumulates context, makes decisions, and executes tasks autonomously over a session.

Where this connects to NordSys

We advise UK businesses on AI-assisted SEO strategy — including how to build content systems that rank across Google and AI search surfaces without continuous manual effort. If you want to understand whether a GLM 5.2 and Hermes setup is appropriate for your current SEO workload, or you need help building a multi-platform content pipeline, we can talk through what makes sense for your situation.

See our SEO & AI Ranking service →