When does it make sense to stop renting your tools and start owning more of them?
With the recent news that a developer used AI to create free open source versions of popular tools like Adobe’s Photoshop and Microsoft Word, we thought it was timely to re-release our episode on open weight models, because the concept is similar. Just like open source tools, open weight models can give your business more control over AI costs, data privacy and the tools you depend on. But what would you give up to get it?
In this week's episode, Jess and Kyle get into what open weight models are, how they differ from closed models like ChatGPT and Claude and from open source, and why the real distinction comes down to access, control and cost rather than anything too technical.
They cover the licensing terms to watch, the three ways to run a model and when fine-tuning a small model for one repeatable job beats a big general model on both cost and accuracy. Open weight models aren't something you need toswitch to tomorrow, but one to have on your radar as AI works its way into more of how your business runs.
Local AI episode: https://earlyadoptr.ai/episodes/local-ai-how-to-run-private-low-cost-ai-on-your-own-computer
"Software is over": Bold AI developer takes aim at Adobe with open source clones: https://arstechnica.com/ai/2026/10/software-is-over-bold-ai-developer-takes-aim-at-adobe-with-open-source-clones/
Llama: https://ai.meta.com/blog/large-language-model-llama-meta-ai/
Qwen: https://qwen.ai/home
Gemma: https://deepmind.google/models/gemma/
Phi: https://huggingface.co/microsoft/phi-4
Groq: https://groq.com/
Together AI: https://www.together.ai/
What You'll Learn
- What open weight models are, and how they differ from both closed models and open source
- The three ways to run one: your own machine, a hosted provider, or your own cloud
- What the licensing terms actually let you do, and when they start to matter
- When fine-tuning a small model for a repeatable task beats a big general model on cost and accuracy
- A simple framework for deciding when open weight is the right call, and when ChatGPT or Claude still wins
Timestamps:
00:43 Why We’re Re-Releasing Our Open Weight AI Models Episode
03:06 Open Source vs Open Weight: How Both Reduce Your Dependence on One Supplier
05:37 Open Source Software and AI: Less Lock-In, More Competition
07:54 Why More Businesses Are Moving to Open Weight AI Models
12:21 Location vs Access: How Open Weight Goes Beyond Local AI
14:54 Why It Matters Now: Reducing Your Dependence on the Big AI Providers
16:54 What an Open Weight Model Actually Is
19:12 How Open Weight Differs From Open Source
24:05 What the Licence Lets You Do With an Open Weight Model
25:29 Why Your Subscription Price Hides the Real Cost of AI
28:32 How Flat Subscriptions Compare to Pay-As-You-Go API Costs
31:03 Control and Privacy Over Where Your Data Goes
34:54 The Three Ways to Run an Open Weight Model
38:18 Using a Model Off the Shelf or Fine-Tuning It
42:22 Llama and Qwen, the Best-Known Open Weight Models
44:02 Emerging AI Models and Their Capabilities
46:34 The Framework for Choosing a Model
51:52 How AT&T Fine-Tuned a Small Model to Sort Call Summaries
53:37 When to Start Considering an Open Weight Model
Resources:
Get in touch with Early Adoptr: hello@earlyadoptr.ai
Follow Us on Socials & Resources:
IG: https://instagram.com/early_adoptr
TikTok: https://tiktok.com/@early_adoptr
YouTube: https://www.youtube.com/@early_adoptr
Substack: https://substack.com/@earlyadoptrpod
Hosted on Acast. See acast.com/privacy for more information.
