scripod.com
Ep 775: Open Source AI 101: Why Local Models, Cheap APIs, and AI Agents Change Everything (Start Here Series Vol 24)

Overview

Shownote

Highlights

Transcript

Chapters

Pins

Ep 775: Open Source AI 101: Why Local Models, Cheap APIs, and AI Agents Change Everything (Start Here Series Vol 24)

Everyday AI Podcast – An AI and ChatGPT Podcast

1 DAYS AGO
Everyday AI Podcast – An AI and ChatGPT Podcast

Everyday AI Podcast – An AI and ChatGPT Podcast

1 DAYS AGO
OverviewShownoteHighlightsTranscriptChaptersPins

Shownote

Until a few months ago, open source AI was kinda a hobby project.  Now, it's tearing corporate boardrooms apart.  Why?  Over the past 6ish months, the gap between frontier closed AI and open sourced AI has shrunk to pretty much nothing. And with the sur...

Highlights

This episode unpacks the rapid transformation of open source AI from a niche experiment to a strategic enterprise priority—driven by unprecedented performance parity, plummeting costs, and new deployment flexibility.
00:00
Chinese model distillation is narrowing the gap between open-source and frontier models
06:48
The gap in capabilities between proprietary and open source models has shrunk by about 90%
09:25
A 10–15 point difference in AI model performance is hard to tell even for those working with LLMs
10:19
Open source models now match the performance of proprietary models from 3–6 months ago and can be run for free on local machines
15:35
Executives are considering providing devices for employees to use always-on agentic AI with local models, which can work autonomously around the clock after a one-time download
17:05
Model distillation is like copying homework, allowing some in China to create cheaper domestic models
22:15
The cost of running a 100-agent swarm drops from over $1200 on OpenAI to about $60 on DeepSeek
24:22
Using open source models means losing legal protection provided by closed, proprietary AI
29:29
Cheaper open-source models are more cost-efficient for high-volume, low-stakes work like summarization and content creation
32:01
Companies can run private workloads on-premise or self-hosted using powerful and efficient open source models
33:33
Premium closed models provide IP indemnification for regulated work

Chapters

Introducing the Firefly AI assistant
00:00
Open source AI cost benefits
03:33
AI model performance differences
09:25
Open source model improvements
10:19
Advancements in local AI capabilities
15:28
Impact of Google's Gemma four
17:04
Introducing Adobe's Firefly AI Assistant
22:15
Adobe Firefly AI assistant beta launch
24:19
Choosing the right AI tools
29:26
Shifting workloads to open source
32:00
Using open-source and closed models
33:31
The future of open models
36:47

Transcript

Jordan Wilson: This is the Everyday AI Show, the everyday podcast where we simplify AI and bring its power to your fingertips. Listen daily for practical advice to boost your career, business, and everyday life. Meet Firefly AI Assistant, now live in Adobe...