Everyday AI Podcast – An AI and ChatGPT Podcast

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

May 12, 2026

AI Summary

5 min read

Open source AI models have closed the gap with proprietary frontier models, making them viable for enterprise use through local deployment, ultra-cheap APIs, and always-on agents, though legal protections are lost.

Capability Gap Closes

Two years ago, open source models lagged far behind closed models from OpenAI, Google, and Anthropic, with a 250-point ELO gap on Arena leaderboards where blind user votes ranked outputs. By late 2025, this shrank 90% to about 30 points—sometimes 15—making differences hard to spot without expertise. Top open source models now match proprietary performance from 3-6 months prior, like end-of-2025 levels from GPT-5.3 or Claude Opus 4.5-4.6. Enterprises previously standardized on one vendor; now boardrooms debate switching for cost savings, potentially millions on tasks like summarization.

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

What you'll learn

1 (00:46) **US Policy on Chinese Distillation** - White House accuses China of distilling US AI models for cheap knockoffs, reshaping open vs closed source debate
2 (01:39) **Open Source Viability in 2026** - Gap between frontier closed models and open source closed dramatically, now only months behind
3 (02:32) **Big Picture Shift** - Enterprises now seriously consider open source over Big Three APIs (OpenAI, Google, Anthropic)
4 (04:29) **Key Takeaways Previewed** - Open/closed default flipped; Gemma 4 local power; payoffs and hidden legal trade-offs
5 (06:21) **From Closed Default to Open Debate** - Pre-2025 enterprises ignored open source; now standardizing on one vendor broken by capability gains
6 (07:22) **Arena ELO Scores Collapse** - Gap shrank 90% from 250 to ~30 points; open source now coin-flip close to frontier
7 (07:57) **Open Source Models Defined** - Downloadable/modifiable under MIT/Apache licenses; run locally for privacy/free use or cheap cloud APIs

+ Full timestamped outline available in the app

Show Notes

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 surge of always on agents driving open models, their development and release schedule is on pace with the frontier labs.

So if your team isn't paying attention to -- and running test cases through -- open AI models, there's a good chance you'll either be overpaying or playing catch up soon.

We walk you through the 101 and what you need to know when it comes to open source AI in this Start Here Series special.

Newsletter: Sign up for our free daily newsletter
More on this Episode: Episode Page
Today's Episode on LinkedIn: Thoughts on this? Join the convo on LinkedIn and connect with other AI leaders.

Upcoming Episodes: Check out the upcoming Everyday AI Livestream lineup
Website: YourEverydayAI.com
Email The Show: [email protected]
Connect with Jordan on LinkedIn

Topics Covered in This Episode:

Open Source AI vs Closed Models Shift
Chinese Model Distillation & Legal Impacts
Enterprise AI Cost Triage Strategies
Google Gemma 4 Local Model Capabilities
Frontier Model Performance Gap Closing
24/7 Agentic AI Systems Overview
API Pricing War: DeepSeek vs US Vendors
Legal Protection Tradeoffs for Open Source AI
AI Workflow Triage: Task-Specific Models
Future Trends: Local and Specialized LLMs

Timestamps:

00:00 Introducing the Firefly AI assistant

03:33 Open source AI cost benefits

09:25 AI model performance differences

10:19 Open source model improvements

15:28 Advancements in local AI capabilities

17:04 Impact of Google's Gemma four

22:15 Introducing Adobe's Firefly AI Assistant

24:19 Adobe Firefly AI assistant beta launch

29:26 Choosing the right AI tools

32:00 Shifting workloads to open source

33:31 Using open-source and closed models