# CLUB 769: AI User Group #19.5: A Hallucination of Agents

**URL:** <https://www.twit.community/t/club-769-ai-user-group-19-5-a-hallucination-of-agents/20482>\
**Category:** CLUB\
**Tags:** episode\
**Created:** [August 29, 2026, 5:00pm UTC](https://www.twit.community/t/club-769-ai-user-group-19-5-a-hallucination-of-agents/20482 "2026-08-29T17:00:11Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Leo](https://sea2.discourse-cdn.com/flex020/user_avatar/www.twit.community/leo/32/16203_2.png) [@Leo](https://www.twit.community/u/Leo)\
**Post date:** [August 29, 2026, 5:00pm UTC](https://www.twit.community/t/club-769-ai-user-group-19-5-a-hallucination-of-agents/20482/1 "2026-08-29T17:00:11Z")

</div>

New Club TWiT episode available: AI User Group #19.5: A Hallucination of Agents

Host: Leo Laporte · Guest: Anthony Nielsen

Topics covered:

- 🤖 Benchmarking local and cloud AI models, including GLM-5.3-Flash, Kimi K3, and Quen 3.8-Flash
- ⚡ Quantization tricks and inference speed comparisons across hardware configurations
- 🖥 Hardware tips: AMD Spark, Mac Studio, and RTX 3090 setups
- 🎙 Voice model selection and assigning personalities to agents
- 🏠 Running a daily AI household: tasks, voices, and memory management

> **[TWiT+ Club Shows: AI User Group #19.5: A Hallucination of Agents | TWiT.TV](https://twit.tv/shows/twit-plus-club-shows/episodes/769)**
>
> Leo Laporte benchmarks the latest local and cloud AI models, including GLM-5.3-Flash, Kimi K3, and Quen 3.8-Flash, across his home agent setup, comparing quantization tricks

#clubtwit #TWiT #Podcast