
Ai builder, I love games. Business inquiries & collaborations: DM or email [email protected]
Auckland, New Zealand
Joined September 2024
- Tweets349
- Following131
- Followers550
- Likes366
Built a Bakusou Kyoudai Let’s & Go!! game with GPT-6 Astra—one of my childhood favorites animate.
Astra Pro’s first browser build blew past my expectations, so I pulled it locally and kept polishing with Astra High.
stack: Three.js
playable: storm-race.vercel.app
post: bubucn.com/en/ai-model-evals…
An English narrative and a substantial graphic upgrade are on the way.
6 weeks ago, PO ZHEN was a demo. Now it’s on the App Store.
Built it with Opus 5.5 and astra, finished strong with fable 5.2
Music & SFX: Stable Audio.
8 Three Kingdoms heroes. 18 chapters, 5 side stories, and fun historical puppet theatre. Free to try.
2
2
210
Opus 5.5 ultracode is unreal.
I just built the game I wanted most as a kid on PS2 — Dynasty Warriors — with Opus 5.5 ultracode. Zhao Yun vs 300 soldiers: combos, dodges, Musou.
Halfway through I wanted it to go faster. Then I opened the preview link and just… stopped. Let it cook.
Play it :bubucn.com/en/ai-model-evals…
Repo:github.com/mike007jd/voxel-m…
stack:three.js
the 5-hour limit hits way harder than the weekly. i KNOW there's usage left, it's right there, i just can't touch it.
The results of this poll has gotten me confused. I definitely thought it would sway much more favorably in the 2X direction across Pro 100/200/500 at the cost of bringing back 5h limits in order to smoothen the demand curve.
120
Set subagents to Opus 5.5 medium and run the main session on Opus 5.5 high. Also set /autocompact to 400k. Powerful and durable.
Just to avoid some confusion: auto-compact does summarize. When it runs, your whole conversation is replaced by a short summary. It doesn't keep the last 1M tokens around.
The ~967K is only when it runs on 1M models. Compacting itself is just one request that reads about as many tokens as the message you sent right before it, and it's mostly cache reads if your cache is still warm.
What @rohit3a found is that waiting until 967K means every message before that point uses a lot of tokens, because each one includes your whole conversation so far. Most of it is read from the cache, which is much cheaper (so this is usually fine as long as your cache is warm), but it still counts toward your usage and adds up as your conversation grows.
/autocompact 400k makes it summarize at 400K instead, so your messages never get that big.
2
179
Used Gpt 6 Astra Pro(site) and three.js build this cyber arm lab demo
Webcam hand tracking to drive it yourself.
Three.js + WebGPU. Still tuning the motion, but the workshop is coming alive.
Playable:bubucn.com/en/ai-model-evals…
Astra is so much faster.....than sol...
Voxel Musou update
New: 7 hero, 4 story chapters, 3 ranked trials.
This trailer made by Claude Opus 5.5.
Play🎮 voxel-musou.vercel.app
Repo📁 github.com/mike007jd/voxel-m…
Voxel Musou updates🏹
• New Hero: Huang Zhong, a bow fighter with aimed shots, fire arrows and a volley Musou
• Story Mode, Ch.1 "Mt. Dingjun"
• Allied troops now fight beside you
• New main screen and character selection page
Playable: bubucn.com/en/ai-model-evals…
Repo:github.com/mike007jd/voxel-m…
Opus 5.5 ultracode is unreal.
I just built the game I wanted most as a kid on PS2 — Dynasty Warriors — with Opus 5.5 ultracode. Zhao Yun vs 300 soldiers: combos, dodges, Musou.
Halfway through I wanted it to go faster. Then I opened the preview link and just… stopped. Let it cook.
Play it :bubucn.com/en/ai-model-evals…
Repo:github.com/mike007jd/voxel-m…
stack:three.js












