writer
Richmond, Virginia
Joined April 2009
- Tweets519K
- Following4.9K
- Followers4.7K
- Likes278K
Pinned Tweet
Some thoughts on the #KennedyCenterClosing
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
Some animator out there is watching this and genuinely rethinking their entire life
Zero hand-drawn frames. Zero video models.
GPT-6.2 Betelgeuse rendered the whole thing purely out of code. Lights, shadows, every movement.
The wild part is it probably finished the entire render before I even decided what to eat for lunch
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
Nice to portray our trajectory in a way that’s not doom and gloom.
Still got that dawg in them
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
🚨 OpenAI hid a message
At 0:36, for 6 frames (~0.2s), the scrambled text settles into: "optimism achieved internally" → O.A.I.
Soft tease that a big model is coming .
And the swirl in that same shot looks like a Bogdanov–Takens point, where a system sits on the edge between stability and sudden change i.e AGI in this context
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
the next phase of ai competition won’t just be about who builds the smartest models, but who can turn that intelligence into products people actually use.
meta is already pursuing agents, smart glasses, mixed reality, and now humanoid robotics.
zuck is going for the whole thing.
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
我严重怀疑 OpenAI 又在憋东西。。。
1. 官号发了个 77 秒动画，一句台词没有，满屏公式：薛定谔方程、麦克斯韦方程，还夹了一行梯度下降。
2. 画面里先写 A PERFECT FUTURE，然后 A PERFECT 被一笔划掉，改成 PEOPLE BUILDING THEIR OWN FUTURES。
3. 据 The Verge，Sam Altman 说 OpenAI 手里没公开的数学发现，比公开的多得多。
宣传片：看完了。
新模型：还没影。
评论区猜一下，先来的是新模型还是新论文？🤔
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
Someone needs to stop Claude Opus 5.5. It turned one picture into a robot octopus I can take apart piece by piece.
i just said "make the model" and what came back floored me.
the build:
> the picture set the look, every petal, lens and link is three.js code
> opus rendered one still next to the picture and fixed the gaps for 3 rounds before it touched the film
> the first pass looked like a toy, so it went back and packed 202 links and 480 wheels into 8 arms
> every arm is one draw call, posed every frame by walking its chain of bends
> something opens every 2-3 seconds: the lens unscrews, the mantle hinges, the pump rises out spinning
> at 0:16 the whole thing blows apart on a 330° turntable and reassembles before the turn ends
25 seconds, 2,616 lines of code and one octopus that leaves the frame on its own.
the skill, the prompt and the live sim 👇
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
Uh oh
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
Mary Phuong of DeepMind on AI loss of control risk.
"The last 2 minutes gave me chills.
A significant amount of people saying that AI companies are paying people to spread fear about these agents.
I have never really felt like [loss of control] was an actual thing to care about.
But seeing Mary and how her face trembles, made me reconsider my stand. She's either a Hollywood-level skilled actress, one of the best, or we should be scared. For real."
"Seems like Mary is already aware of more than she can admit and it haunts her..."
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
I’m convinced this entire video was created by GPT-6.2 Betelgeuse, entirely through code. No video generation model. Just AI writing the code to bring every frame to life.
This is absolutely insane. We’re witnessing the future of AI-generated content.
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
GEMINI 4 CARBON BURIED OPUS AND ASTRA THIS HALLOWEEN 💀
Same prompt: a zombie hand claws out of a grave and rots down to the bone.
· Gemini 4 Carbon: 1h 23m · $10.2
· Opus 5.5: 1h 6m · $8.1
· GPT-6 Astra: 24m · $5.7
Carbon went full horror movie: candlelit graveyard, labeled anatomy at every stage.
Opus nailed the rot but lost on the scene.
Astra finished first and built a landing page with a hand on it.
Halloween test via aimlapi.com
GOOGLE'S NEW GEMINI 4 CARBON IS SCARY GOOD AT 3D 🥶
One prompt: a robotic eye that blinks, focuses, then explodes into every part.
Gemini 4 Carbon → 29m 49s
Waiting for it on aimlapi.com
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
5 weeks in an Opus 5.5 is still the goat.
It has the magic and feel of my beloved Opus 4.6, the intelligence of Fable 5.5, and the token efficiency of GPT-6.1 Sol.
5 weeks after its release, Opus 5.5 remains by far the best model overall.
Therefore, I expect a major release from OpenAI next week, because it's clear that Anthropic will also be releasing Fable 5.5 next week, and most likely a new Opus as well.
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
🚨 Gemini Carbon 4 genuinely is so good in threejs
this is a google model we are praising , no laziness at all
now it think like really thinks before output sometimes 8-10 min
This is seriously going to be a great model , i just hope now google fix webdev and slop ui , buy data from your previous supplier google
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
Models are cooked
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
🚨 Gemini 4 Carbon: New Checkpoint
> the same gemini-4-argon checkpoint on arena has been updated to '01a161e8-cebc-7960-a363-fc6a0e8ab276'
> Latest checkpoint observed as of October 30, 2026, at 6:29 PM PT
Full details in the comments
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
🚨 OPENAI MAY HAVE ALREADY ACHIEVED GPT-6.5 ASTRA INTERNALLY.
Here's what's fueling the speculation:
→ OpenAI recently showcased a fully animated video generated using nothing but code.
→ The results look incredibly impressive, potentially challenging what we've seen from Claude Opus 5.5.
→ An advanced Astra checkpoint could already exist internally.
→ There's speculation about a possible launch as early as today.
But here's the catch:
→ Potentially expensive inference.
→ Could be exclusive to the $500 subscription tier.
Imagine having an AI model capable of generating complete animated experiences through code.
Is OpenAI preparing its next big move against Anthropic?
Nothing is confirmed, but the speculation is getting interesting.
Shelton Bumgarner (sheltgarner.bsky.social) retweeted
突发 突发！！！
GPT-6.2 Bel 即将发布
据消息人士透露，OpenAI 报告的 Bel 实际上是 GPT-6.2 Betelgeuse。
是超大参数量的全新基础模型，该模型拥有 12 万亿参数，除此之外在视频表现方面也表现不俗，看样是是要跟Opus5.5硬钢一次了。
一位经过验证的 OpenAI 内部人士声称，它在 Artificial Analysis 上得分 72，比 Carbon 高 13 分。
我找了半天居然没有看到Opus5.5的身影
不管分数多高，没有Opus5.5的评分榜始终是不完整的。
