Try Recraft V4.1 now and experience an image generation model with exceptional visual taste :
PewDiePie built an open source AI workspace called Project Odysseus and I put it through its paces so you don’t have to.
Odysseus is a self-hosted AI platform that lets you run local AI models, connect APIs, do deep research, manage files, and keep your data completely off the cloud. Think ChatGPT or Claude but running on your own machine, with your own models, and full privacy control.
In this video I install it on my M3 Mac, connect local models through Ollama, run model comparisons between Gemma 3 12B and GPT 5.5, test the deep research module, and explore every feature the app has to offer including the brain memory system, the model comparison scoreboard, and the image editor (spoiler: some things work great, some things I absolutely could not get to work no matter how hard I tried).
If you’ve ever wanted to run AI completely offline without paying API fees or sending your data to a server, this is worth knowing about.
Discover More:
🛠️ Explore AI Tools & News:
📰 Weekly Newsletter:
Socials:
❌ Twiter/X:
🖼️ Instagram:
🧵 Threads:
🟦 LinkedIn:
👍 Facebook:
Resources From Today’s Video:
Let’s work together!
– Brand, sponsorship & business inquiries: [email protected]
#AITools #LocalAI #PewDiePie
Time Stamps:
00:00 PewDiePie’s Secret AI Project
00:23 What Odysseus Actually Is
01:39 GitHub Setup & Install
04:18 Sponsor: Recraft
06:12 Sidebar Tour
07:04 Connecting Ollama & Local Models
08:13 Testing The Local Model
09:50 Adding OpenAI’s API
10:41 Trying A Bigger Local Model
11:18 Brain Memory
11:49 Model Compare & SVG Test
16:39 Deep Research Test
21:01 Scoreboard Results
23:01 Calendar, Gallery & Notes Tour
23:54 Fighting The Image Editor
27:13 What Worked, What Didn’t
29:43 The Future Of Local AI
31:00 Final Thoughts
source
