There's a new best OSS model and it's...weird
Related videos
We all fell for it…
Theo - t3․gg
216.3k views
This model is kind of a disaster.
Theo - t3․gg
166.2k views
It's finally here.
Theo - t3․gg
152.2k views
Opus 4.6 Is The Best Coding Model Ever Made*
Theo - t3․gg
106.6k views
The Best Model For Frontend Design Is...
Theo - t3․gg
124.4k views
Kimi K2.5 might be my new favorite model...
Theo - t3․gg
95.3k views
GPT-5.2 is the best model ever made*
Theo - t3․gg
100.4k views
Anthropic won. This is my new favorite model (sorry Gemini…)
Theo - t3․gg
101.3k views
Gemini 3 Pro is the best model ever made
Theo - t3․gg
163.6k views
Is gpt-5.1 the best code model ever?
Theo - t3․gg
63.2k views
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Unlock all features
FREE: Get instant access to 10 AI summaries, chats, or transcripts per day.
Related videos
We all fell for it…
Theo - t3․gg
216.3k views
This model is kind of a disaster.
Theo - t3․gg
166.2k views
It's finally here.
Theo - t3․gg
152.2k views
Opus 4.6 Is The Best Coding Model Ever Made*
Theo - t3․gg
106.6k views
The Best Model For Frontend Design Is...
Theo - t3․gg
124.4k views
Kimi K2.5 might be my new favorite model...
Theo - t3․gg
95.3k views
GPT-5.2 is the best model ever made*
Theo - t3․gg
100.4k views
Anthropic won. This is my new favorite model (sorry Gemini…)
Theo - t3․gg
101.3k views
Gemini 3 Pro is the best model ever made
Theo - t3․gg
163.6k views
Is gpt-5.1 the best code model ever?
Theo - t3․gg
63.2k views
Top Comments (10)
WAIT -- "Wait's Average Index by Theo"
so, it's essentially that Punisher meme where he goes "wait!wait!wait! WAITWAITWAITWAITWAIT!!"
i swear i saw this model like 2 weeks ago
I'm not hyped about LLMs, but your videos are very helpful. You're saving us time. Otherwise, we'd have to go through every other LLM ourselves and see what it offers. Thank you!
Video seems a bit late but I thoroughly enjoyed your breakdown regardless. I hope you continue making all these kinds of videos, you’re one of my go-to-sources for programming news lol
I played a drinking game, where every time Theo says interesting I take a drink. I ended up in the hospital.
19:50 I love how AI predicted a human taking over the chat and complaining that the code is wrong.
We missed rotating hexagon with red ball in it benchmark
The "hmm/wait" benchmark is amazing.
Its tricky to setup I've found. The biggest tricks are to set repetition penalty to 1 because it is a reasoning model it can repeat things while thinking if the penalty is applied the reasoning fails. Second put the <think> tag with new line in the system prompt or instruct template. Tells the model to only start thinking after the context.
Unlock the Data Inside
Turn Videos into Knowledge
- Get FREE 10/day: transcripts, summaries, chats
- Chat with videos, export text & PDF
- $1 free API credit for RAG, chatbots & research
Free forever plan • All features unlocked
Top Comments (10)
WAIT -- "Wait's Average Index by Theo"
so, it's essentially that Punisher meme where he goes "wait!wait!wait! WAITWAITWAITWAITWAIT!!"
i swear i saw this model like 2 weeks ago
I'm not hyped about LLMs, but your videos are very helpful. You're saving us time. Otherwise, we'd have to go through every other LLM ourselves and see what it offers. Thank you!
Video seems a bit late but I thoroughly enjoyed your breakdown regardless. I hope you continue making all these kinds of videos, you’re one of my go-to-sources for programming news lol
I played a drinking game, where every time Theo says interesting I take a drink. I ended up in the hospital.
19:50 I love how AI predicted a human taking over the chat and complaining that the code is wrong.
We missed rotating hexagon with red ball in it benchmark
The "hmm/wait" benchmark is amazing.
Its tricky to setup I've found. The biggest tricks are to set repetition penalty to 1 because it is a reasoning model it can repeat things while thinking if the penalty is applied the reasoning fails. Second put the <think> tag with new line in the system prompt or instruct template. Tells the model to only start thinking after the context.