GLM-5: From Vibe Coding to Agentic Engineering

2026-02-11 · Source: Simon Willison's Weblog · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, quick

Summary

Z.ai has released GLM-5, a new MIT-licensed large language model featuring 754B parameters and a 1.51TB footprint on Hugging Face. This model is twice the size of its predecessor, GLM-4.7, which had 368B parameters and was 717GB. The release also highlights the emerging term "Agentic Engineering" to describe professional software engineers who build with LLMs, a term also noted by figures like Andrej Karpathy and Addy Osmani. An example prompt, "Generate an SVG of a pelican riding a bicycle," run through GLM-5 via OpenRouter, produced a high-quality pelican but a less satisfactory bicycle frame, as observed on February 11th, 2026.

Key takeaway

For Computer Vision Engineers evaluating new large language models, you should consider GLM-5's substantial parameter count (754B) and MIT license for potential integration. Be prepared to assess its output quality for specific elements, as demonstrated by the pelican/bicycle example, to ensure it meets your project's detailed requirements.

Key insights

GLM-5 is a significantly larger MIT-licensed LLM, underscoring the trend towards "Agentic Engineering."

Principles

• Larger models may not guarantee uniform quality across all output elements.

In practice

• Experiment with GLM-5 for SVG generation tasks.
• Evaluate specific output components for quality variations.

Topics

• GLM-5
• Large Language Models
• Agentic Engineering
• Model Parameters

Best for: Computer Vision Engineer, AI Engineer, Machine Learning Engineer, Prompt Engineer

Related on AIssential

• A Review of GLM-5: 744B-Parameter LLM Built for 200K Context and Agentic Training — GLM-5 bridges prompt-driven coding to agentic engineering via systems-level LLM advancements.
• Insane Open Source AI Model Just Dropped — GLM 5.1 is an open-source model matching proprietary LLMs in software engineering benchmarks.
• GLM-5.2 First Test | Frontier-Level Open Weight Model? | Agentic Coding, Web Design, Logic Puzzles — GLM 5.2 is a frontier-level open-weight model excelling in agentic coding and complex web design tasks.
• NEW GLM-5 vs MiniMax-2.5: NEW = BETTER? — New large language models struggle with complex causal reasoning puzzles despite advanced architectures.
• The Sequence AI of the Week #813: Deep Diving Into the Amazing GLM-5 — AI is shifting to autonomous "agentic engineering" requiring advanced reasoning and context handling.
• GLM-5.2: A Free AI Just Beat GPT-5.5 at Coding. Here’s What the Headlines Left Out. — Z.ai's GLM-5.2 offers a 1M-token context and MIT license, outperforming GPT-5.5 in coding.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Simon Willison's Weblog.