FindLogan

SWE who won't shut up about AI agents.

Follow along to learn more about AI👇

[FindLogan @FindLogan]

I got early access to Seedance 2.0. from a friend. The results are stunning.

The US should be getting it soon, but in case you guys are into Anime (and if you're not, I think this is still a good watch).

What are you going to create when it comes out to the US?

[FindLogan @FindLogan]

@thsottiaux "Sorry we had a problem"
"What are you going to do about it??"
"Heres a token reset!!"

[FindLogan @FindLogan]

@dedene Literally any other tool is cheaper. I made a 2 subagent request. Cost over $4 for just searching the codebase... just impressive.

[FindLogan @FindLogan]

Grok is one of those tools that are quite explosive. One moment, I toss them aside for complete irrelevancy in software, then the next moment - they're reputable and coming with vengeance.

Truly great work they're doing over there on the Grok team! Looking forward to the day it's one of my go-to tools.

[FindLogan @FindLogan]

@thsottiaux I have a relative who has codex'd an entire device. Including advice on soldering, wiring, and even putting together a 3d printed case. Truly remarkable how much it's able to guide people through in terms of physical devices.

[FindLogan @FindLogan]

@sama I find it crazy how nuts everyone goes when Sam quotes the bible.

I find this purely inspiring that even people in Sam's seat are not ashamed to talk about this.

[FindLogan @FindLogan]

@rezoundous I'm sure this is potentially a case of them throwing compute at a new model if it's tweaks like all are saying.

Haven't tried it yet, but cautiously optimistic. Compute is compute though..

[FindLogan @FindLogan]

@droidbuilds Please. Sign me up! I want to participate... easy money.

[FindLogan @FindLogan]

@ai_for_success Kimi may have not been the fastest model.. but it was most definitely reputable!

[FindLogan @FindLogan]

@jasondoesstuff It would be nice to see a benchmark for this. See what the degraded benches are when people are complaining.

[FindLogan @FindLogan]

@jun_song Claude also has a ton of outages and service degraded events. Not surprised one bit.

[FindLogan @FindLogan]

@steipete The comments on his post outlines that it is cursor. There was someone who reach out to him about it.

Still pretty insane though!

[FindLogan @FindLogan]

@elonmusk The deal for X is rapidly improving. Grok may eventually take the lead with agents if this continues much longer.

[FindLogan @FindLogan]

I outlined DeepSeek's capabilities in a previous post. Huge news.

Very capable model.. they lag behind GPT-5.4 and Gemini 3.1 but this is a great budget model!!

Good deal if youre on a budget 🤝

[DeepSeek @deepseek_ai]

We are making our discount permanent! 🎉

Enjoy building with DeepSeek-V4-Pro and bring your innovative ideas to life! 🚀

[FindLogan @FindLogan]

Goal mode for codex is out.

OpenAI is lowering the bar for your ambitious vibe codes projects!

With GPT-5.5 being the most capable model but lacking in design still you can probably expect developer quality apps. (Meaning the ui might not be pretty but at least the app functions!)

Gemini is still my first grab for design choice. Toss the mock to codex and let it grind.

What are you building right now? 👇👇

[OpenAI Developers @OpenAIDevs]

🥅

/goal has graduated from an experiment���for tasks big and small, Codex gets your work done.

Use goal mode in the Codex app, IDE Extension, or CLI to give Codex a specific milestone, and it will keep working until it gets there, even across hours or days.

You can check in and steer, and even pause Codex along the way.

Pro tip: start side chats to understand the work that has been done so far without having to interrupt the main task.

https://t.co/e55oI6AhYM

[FindLogan @FindLogan]

@OpenAIDevs Everyone jokes about vibe coding using Grok in a Tesla... guess OpenAI wanted their fair share of the vibe code & drive communities.

Jokes aside... amazing feature. Great job team!

[FindLogan @FindLogan]

@OpenAI @kenjihata @adele__li @AndrewMayne Im not remotely surprised. ChatGPT holds the BEST image Gen on the market to date arguably.

[FindLogan @FindLogan]

@OpenRouter @GoogleDeepMind Solid company, always fast to add models. ⚡️

Looking forward to benchmarking these!

[FindLogan @FindLogan]

@sama Imagine this post lining up with the weekly reset.

[FindLogan @FindLogan]

Gemini 3.5 Flash is here.

> 1-Prompt designs aren't too bad!
> Lots of promises from Google
> Visual Benchmark is looking promising, but needs some more testing
> They claim it is "12x faster on Antigravity 2.0 instead of just 4x"

What do you think? 👇👇

[Google Gemini @GeminiApp]

We announced a lot of Gemini updates at #GoogleIO, including:

- A complete redesign of the Gemini experience with Neural Expressive
- Agentic experiences coming to Gemini with Daily Brief & Gemini Spark
- Gemini Omni & 3.5 Flash models
- and more!

Catch up on everything here 🧵

[FindLogan @FindLogan]

@AiBattle_ Hopefully we'll get Gemini 3.2 inside the console as well? Or even just a straight up 3.5?