Rez

Just getting started on X & YouTube. Learning as I go.
YouTube:

[Rez @RezAIT00]

@lyraxana Is it just me or are these generations kind of mid?

[Rez @RezAIT00]

@LDT0545 UI

[Rez @RezAIT00]

@TitanidesLeto It would be amazing if it was this week

[Rez @RezAIT00]

@Ananth7e I am not convinced that a bunch of bots didn’t vote

[Rez @RezAIT00]

@theo I like it, good job.

[Rez @RezAIT00]

@WescheNex1q I think it can do really well with a opus orchestrator.

[Rez @RezAIT00]

@giadotai You are a genius.

[Rez @RezAIT00]

I could not wait to show this. It took 4 hours, it used 1 opus agent, and ~10 haiku agents. This was mostly made using haiku.

It is weird to think a SMALL model can generate this level of quality.

Music/Sound: Eleven Labs

[Rez @RezAIT00]

I have been testing agent swarms using haiku for agents and opus as the orchestrator. I have to say the generations you can get are amazing everyone should try this. I’ll show some generations tomorrow.

[Rez @RezAIT00]

@ChrisGPT I don't see how any games will have a "pre-release" they will just get copied and released early.

[Rez @RezAIT00]

@ItsmeAjayKV Imagine Qwen 4-27B next week....A man can dream.

[Rez @RezAIT00]

@alexgetmancom I hope openai makes a big come back. I don't see myself renewing anytime soon.

[Rez @RezAIT00]

@TokenGremlin Are you happy with this? I think this deserves another reset.

[Rez @RezAIT00]

@TokenGremlin The human review part is what gets me. If AI can produce months of work in a day, but checking it still takes months, how much of that speed can we actually use? Verification could end up being the real bottleneck.

[Rez @RezAIT00]

I have been testing agent swarms using haiku for agents and opus as the orchestrator. I have to say the generations you can get are amazing everyone should try this. I’ll show some generations tomorrow.

[Rez @RezAIT00]

@theo T3 is great Theo, but I wish we could have groups in threads. So one thread could own 3 different chats for example.

[Rez @RezAIT00]

@Da7_Tech I wonder where we’ll be in a year from now… will we call that AGI? I remember people calling o1 AGI now we have local models that can run at a level higher than that… crazy times

[Rez @RezAIT00]

@MatthewBerman The only thing I hate about local is that it’s so hard to run “multiple agents” at a useful context

[Rez @RezAIT00]

@notjazii I had a thought the other day: We aren’t close to the “ai” we see in the movies. The kind that can calculate the size of something with a glance or the distance of an object relative from where you are. Etc.. I wonder how many years we are away from that.