oh Chadol

Building an AI tool to identify products from videos.

Sharing the journey from idea to launch.

[oh Chadol @OhChadol]

Internal situation at Anthropic

[oh Chadol @OhChadol]

Tibo.jev : reset (0.77)

[oh Chadol @OhChadol]

Yuanlei Niu says he built a playable 2.5D Splatoon-style prototype with GPT-6 Astra and Blender. He says Astra helped him iterate the layout, materials, lighting, textures and composition in Blender. That could be useful for moving from concept to a playable build.

[Yuanlei Niu @truenyl]

I built a 2.5D Splatoon-style game with GPT-6 Astra + Blender.

This one turned out way better than I expected.

I used some assets AI found online, then let GPT-6 Astra work inside Blender to keep improving the layout, materials, lighting, textures and overall composition.

The biggest difference compared with my earlier 3D experiments is that Blender gives the whole thing much more visual polish.

And the gameplay is actually in pretty good shape too.

The shooting feels surprisingly solid, there’s a decent variety of weapons, and the AI enemies already have some numerical difficulty differences instead of all behaving the same.

The game is playable now, and I deployed a build on Vercel so people can try it directly.

https://t.co/h3131SieyB

Now I really want to add multiplayer hhhhh.

[oh Chadol @OhChadol]

Suddenly Opus 5.5? Did Anthropic get really rattled by falling behind OpenAI?

[Pankaj Kumar @pankajkumar_dev]

Opus 5.5 Coming This Tuesday

- Opus 5.5 is in stealth testing under the codename claude-wafer-eap. It was previously to be Opus 5.2, but now they have changed their plan to Opus 5.5.
- I think they changed it because Opus 5.5 has significant changes compared with Opus 5.1.
- Its around the level of Astra and Fable 5.1, and in some areas can perform even better.
- Its quite good at 3D and frontend work, with better attention to small details and more polished outputs.
- Pricing: $4 input / $20 output per 1M tokens, $0.20 cache read / $5 cache write. Context window: 872K down from 936K.
- This makes it around 20% cheaper than Opus 5.1.

And this time, Opus 5.5 > GPT-6 Sol. Anthropic gonna be back.

[oh Chadol @OhChadol]

@doymun 영미권은 ai slop에 대한 거부권이 심해서인지 전혀 사람들이 안보더라고요...

[oh Chadol @OhChadol]

@doymun 한국어만 하고 계신가요?? 저는 영어를 했는데.. 영어는 그냥 안되네요...

[oh Chadol @OhChadol]

@clipscircuitHQ @doymun 알려주긴 어렵네요... 그냥 심플합니다

[oh Chadol @OhChadol]

Pippit 3D Director Studio lets you block out a shot before making the ad. Set the camera and scene elements in 3D, capture a reference, then carry it into Canvas for AI images or video. Could be useful for paid promo work.

[Pippitofficial @pippitofficial]

Your character won’t move right.
Your camera has other plans.
Your credits? Gone. 💸

Meet Pippit’s 3D Director Studio — take control of your characters, camera, and scenes instead of hoping the next generation gets it right.

You’re the director now. 🎬

#Pippit #PippitAI #3DDirectorStudio #Seedance25 #creativeagent

[oh Chadol @OhChadol]

@doymun 저는 mcp로 chatgpt로 이미지 7장 만들고 codex로 3장 만들고를 반복하면서 만들고 있습니다

[oh Chadol @OhChadol]

It seems Thiebaud promised a reset. The ending is a bit ambiguous, though. Is he launching a product on Tuesday? Or is he doing a reset? I think it's a reset.

[oh Chadol @OhChadol]

@thsottiaux @udiWertheimer So, does that mean a reset on Tuesday? Your words are quite ambiguous.

[oh Chadol @OhChadol]

This is useful if you're already spending $200 but still constantly running out of tokens and need a way to get more usage.

To give an extreme example:

Say I subscribe to 5x for $100 and use up the entire 5x allowance in just one day.

Then I upgrade to 20x that same day for roughly another $100.

I've paid about $200 total ($100 + $100), but I've already used the full 5x allowance AND I still have 100% of my new 20x allowance available.

That's the key idea: use up 5x first → upgrade → get a fresh 20x allowance.

[oh Chadol @OhChadol]

Most people don’t know this:

You can use up your entire 5x limit — and then reset your usage by upgrading to 20x.

I’ve been using this to effectively get an extra usage reset about every two months without staying on 20x all the time.

Here’s what I do:

Jan 1 → Start on 5x
Jan 15 → Use up the 5x limit
→ Upgrade to 20x → Usage resets

Feb 15 → Switch back to 5x
→ Use it up → Upgrade to 20x again → Reset again

So the cycle is simple:

5x → Use it up → 20x → RESET → 5x → Repeat

If you burn through your 5x limit quickly, this can give you more total usage for your money than simply staying on 20x from the beginning.

[oh Chadol @OhChadol]

ChatGPT usage has been expanded 20x. Subscribe now!

[oh Chadol @OhChadol]

I finally got approval for Jev, which has been all the rage lately.
But... what should I try doing with it? Should I have it trade Bitcoin?
Maybe use it to sort emails?
Please give me some ideas.

[oh Chadol @OhChadol]

This is pretty surprising.

The SVG output is way better than I expected. LuminaBench labels this a Gemini 4 Pro PS5 example, and says it took around 10 minutes to finish.

The model label still is not confirmed, but the quality here is hard to ignore.

https://t.co/w4f74j9wWM

[Lumina @LuminaBench]

🚨 Gemini 4 Pro is crazy (PS5 SVG example)

This might genuinely be one of the craziest outputs I’ve got from an AI model yet

For context this took around 10 minutes to complete too

We are in for a treat when this thing fully drops

[oh Chadol @OhChadol]

Interesting early sample, but not a benchmark yet.

The post labels the checkpoint Gemini 4 Pro while saying it is appearing in Arena under “Gemini 3.8 Flash.” On one pelican SVG prompt, it looks surprisingly close to GPT-6 Astra Pro.

Promising signal, not a model verdict.

[Lumina @LuminaBench]

🚨 Gemini 4 Pro vs GPT-6 Astra Pro (Pelican Riding a Bicycle SVG)

Both look quite similar, maybe Astra is slightly better imo

But remember this is still a Gemini 4 Pro checkpoint (argon-d), so it’ll probably improve from here

Very bullish on Google rn after testing this out