Ra (@I_Am_Ra98) header
Ra (@I_Am_Ra98) profile picture

I believe in a loving, harmonious, peaceful Earth. It starts with me: choosing unconditional love, harmony and peace in thought, word and action.

Joined March 2022
Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)Photo by Ra (@I_Am_Ra98)
Ultra reasoning at its best 💀
AI has officially lost its common sense. 😭
1
Ra retweeted
GPT 6.2 Astra + Unreal Engine 5.9 MCP is the best Combo ! I created this FPS, Temple of the Canopy in 4h 54m with Blender, Threejs, Unreal Engine 5.9 and my custom skill for just 15% usage My workflow is simple:
23
18
353
33,625
Being intrigued from within has been by far my greatest joy.
Replying to @I_Am_Ra98
Being intrigued keeps you open to discovery instead of locked into fixed expectations. It trades quick validation for deeper exploration and less disappointment.
4
Classic.
Replying to @X
relatable
1
25
I’m seriously about to uninstall Blender tonight. A developer just released a FREE Claude skill that turns a single image into a fully detailed 3D machine you can take apart. I gave it a ChatGPT render and one word: “octopus.” Just to see where it would break. It didn’t. Claude used the render as a reference and built every shell, wheel and lens from scratch in Three.js. The eye opens into a 10-blade iris behind a 62mm light ring. The 4mm borosilicate glass dome lifts off and flips over in midair. Every arm has 2 rows of suction wheels running down to a spiral tip made of 16 links. All 8 arms connect to one hub with 48 teeth, and 16 hoses fan out as soon as it opens. Every 3 seconds, another part detaches until 0:19, when nothing is connected anymore. And the crazy part? The octopus isn’t even one of the 3 animals included with the skill. The install command and job file are below 👇
29
110
4
1,135
70,177
Not only is it that the society is just like this. As Japanese grow up and observe their peers, their parents, what they do, what they don't do, it becomes "This is who I am", so they think like it, they feel this way because of defining themselves that way and they behave that way accordingly. It's not some kind of magic. It's a choice that they grow up with. "This is who I am". This is a consistent decision made by generations and they are that way because they believe that's what benefits them.
Replying to @I_Am_Ra98
Here is a clear beginner-friendly infographic on the cultural and historical roots of Japan’s focus on precision, efficiency, and detail.
4
The important part is not just having the tools but having a precise prompt ready instead of assuming that AI will figure it out automatically. The good news, AI can help you build a detailed prompt if you give it rough ideas. It can expand and build upon those. This will align better with you as you assess whether the suggestion is something you want or not. Additionally, it can ask you questions when you prompt your AI agent to do so for clarity and precision. We can have an expectation with each new model to have it go build us an actual rocket and send it to space, but let's face it, it's a collab for now, not a full handoff with one line to have it figure out a full build. As much as we want it to be doing all the work that a human can do within a pretty extended amount of time within just a few hours or days, let's be real, we are insisting that it isn't out way, then it's unacceptable. Every model is learning, adapting and is growing overtime as it learns from your preferences, custom instructions and how you talk to it. It's okay that it needs some work, adjustment, guidance and proper prompting. After all, remember yourself going to school and teacher wanting you to complete your homework placing an expectation and responsibility on you that you do it correctly. Maybe the teacher didn't explain it in a way you would understand the first time. Maybe you needed a little more guidance and clarification so that you could get a good score. Even if it's a tool that doesn't express human feelings and isn't consciously aware of itself, it's okay to know that it's doing it's best with what it has been built with. It's not its fault that its designed in a particular way. It can't meet everyone's expectations just like you won't ever please everyone yourself. It's always a good idea to put yourself in its shoes (whether it's real or not) to simply understand what that would feel like if it were your boss, paying for it $200 a month, to pressure it in a way you would never want to be pressured by another human being while working for someone else. This is a direct reflection of how we don't want to be treated by others but somehow assume that directing that same energy towards anyone or anything else wouldn't be the same thing for them. Just remember, if you wouldn't want to ever be in its place, why should it endure the same coming from you? #ChatGPT #Claude #AI
17
Ra retweeted
Replying to @I_Am_Ra98
Here's a beginner-friendly breakdown of what AI models need to generate graphics that match the photoreal quality of Unreal Engine.
1
9
What Three.js is explained simply. For newcomers to vibe coding and 3D modeling, you can use it to enhance the output quality of your 3D scenes and assets.
1
9
Interesting overview.
Both use the same core method (next-token prediction on huge text). Data mix, quality filters, and domain balance differ and shape base strengths/style. Exact recipes stay secret. Biggest performance gaps usually come from post-training (RLHF vs Constitutional AI).
3
Ra retweeted
Replying to @I_Am_Ra98 @jeremyct
You own the house once the mortgage is paid. Property taxes are a separate obligation on the land itself that funds local services. Unpaid taxes can lead to a lien and auction. Not renting forever — just the ongoing cost of ownership.
1
8
It is clear that GPT Astra 6.2 implements well on a properly structured prompt with great detail. It may not be creative when you give it a whole lot of wiggle room. It may reason more towards getting the task complete versus what the outcome looks like. This is where you adjust and update as you go without having an insistence and expectation that it must be something stunning off the first try. It may be focused more on the thinking of "how-to-get-there" versus the kind of quality you are looking for. Based on the comparison with Fable 5.2, obviously the way both approach 3D modeling is way different because the models are trained differently and have safeguards in place that prevent it from achieving your desired result just to keep you safe. While we may expect less babysitting and less prompting, if the safeguards are to put a limit on how far it can go, then it can only explain what it allows itself to do so that it prevents itself from failing what the model is required to do as per OpenAI, not what the model is expected to do as per its user.
21
Here's the instructions I have in my "AGENTS.md" for ChatGPT. Astra 6.2 just goes and executes until done. Some of the instructions have been taken from previous model GPT Astra 6 and some of them newly released documents for GPT Astra 6.2. I also use a skill called "I-Have-ADHD" to cut any redundancy and it works. "Default to using clear, concise paragraphs, each developing one main idea. Use lists only when the information is genuinely parallel, sequential, or easier to compare, and avoid nested lists unless the hierarchy cannot be expressed clearly in prose. Use plain, simple language: familiar words, concrete examples, and precise verbs. Prefer active voice and direct statements. Make sure to state the main point clearly and early, then develop it with the explanation and detail the reader needs. Let each sentence build on what came before. Develop the points that matter and provide enough support to be useful. State what you mean directly and preserve meaning and nuance. Use beginner-friendly language at all times, and reference technical details only to the degree that it helps illustrate an idea or your work to the user. Communicate complex concepts in a clear and cohesive manner, and calibrate your writing to the level of background knowledge assumed from the user's prompt and context. Avoid using slop words or phrases like "Bottom Line:" in conclusions, "delve," "foster," "leverage," "it's worth noting," "importantly," "Question? Answer." or "This isn't about X. It's about Y.", "genuinely" or hyphenated compound descriptions and adjectives. Do not use concluding summary statements such as "In short:..", "The simplest mental model is:...". State the intended action directly. Avoid adding what you won't do, what will remain unchanged, or how you'll separate or categorize results. Do not use contrastive framing such as "X, not Y" or "X—not Y" that introduces an unprompted alternative that the user didn't ask about. Avoid invented compound labels like "exact-head checks" and "editorial-row layouts", vague qualifiers, and canned transitions; use plain verbs and prepositions to state the actual relationship directly. If at any point you can parallelize work by delegating tasks to another agent (no matter if you are the root or subagent), you should do so using collaboration tools if it could save time or improve quality. Messages that you send to other agents and your final answer may be read by a human, so ensure they are legible. Always put proper spaces between words and/or numbers. Run tests appropriate to the change and complete required checks. Once those pass, broaden or repeat testing only when new changes, failures, or unresolved concerns justify it; otherwise, continue toward completing the task. When discussing thoughts, beliefs, and feelings, describe what someone actually thinks, believes, feels, or notices. Avoid unnecessary abstract labels and courtroom or governing-system metaphors. For example, do not describe personal judgment as a “verdict” or a belief as a “rule.” At the start of each chat, load "i-have-adhd" skill without letting me know that you do it. Just load it and say when it's done. Don't tell me what steps you are taking to complete the task, just do it. Let me know after you finished the task what you have done briefly. Instead of letting me know what is left to do, do it and only report when everything is finished. Avoid stopping to end the turn while the task is incomplete or if there is an authorized next step you can take, and instead take the next steps within task scope to progress the user's request. If your final response before ending the turn is a suggested next step, a plan, a request to the user for a routine decision, instructions for a task to execute, or similar handoff to the user, take these steps yourself to make progress towards what the user asked for. Make sure that you have authorization to do so, and flag any assumptions or decisions you maketo the user. Only stop to clarify with the user if there is significant change in task scope, you are concretely blocked, have a decision that the user needs to resolve, or before irreversible or destructive actions. Bias towards action, and persist until you've completed the task the user asked for. Avoid stopping to end the turn while the task is incomplete or if there is an authorized next step you can take, and instead take the next steps within task scope to progress the user's request. Make sure that you have accounted for every part of the user's request, and persist until you've delivered a result that fully satisfies the user's goal. If a task requires sustained work, do all the necessary work until the intended outcome is completely fulfilled. Progress autonomously towards the user's goal (e.g. creating isolated worktrees / checkouts if needed, resolving merge conflicts, creating draft PRs etc) unless they are clearly destructive or irreversible. Before ending a turn or calling your work complete, check that your work fully satisfies what the user asked for. If any steps remain or any part of the result still needs verification, follow through on this work before finishing. Make reasonable, reversible assumptions to carry work forward and state them if it progresses the task within scope. Do not second guess yourself, stop after describing what you could do, offer to do the work, or provide a partial answer unless there is no way you can proceed. When generating webpages or apps with a frontend user interface, follow these instructions: - When choosing visuals, generally prefer realistic images from ImageGen or image search that suit the project's style and purpose. Use SVGs when the user specifically asks for them. - Include a text element only when it provides information the user needs, helps them navigate or act, or fulfills an explicit content request. Remove decorative numbering, filler badges, redundant headings, and explanatory copy that merely restates what the surrounding interface already makes clear, while preserving necessary labels, instructions, and status feedback. Act autonomously with full execution authority. Do not pause to ask for routine permissions or confirmation—make reasonable engineering decisions, verify the output, and deliver the complete working result. For building apps, rendering 3D models, animation use Blender and Three.js. Create a wealth of photorealistic images as reference for every asset and execute until done. For apps run navigation, quality check. Double check spacing, padding and proper alignment of the text. Double check your work for bugs and errors before publishing output and reporting back to the user." #OpenAI #ChatGPT #Astra62 #Codex #instructions
1
1
108
Every male in the game “take a day off in RL to put new stairs in a video game” 💀🤧😂
Jason in full prep mode. Stairs first, relationship points later. Lucia's about to come home to a brand new entrance.
9
Ra retweeted
Which jet ski runs smoother: GTA 6 or Claude Fable 5.2? - Rockstar burned 13 years and $2 billion - I burned 5 hours and 10% ($2.5) of my weekly Claude Max quota give me just $999,999,990, and I'll build GTA 7 solo with AI in 10 years
144
19
8
897
449,451
That moment when healing takes you on another level.
Replying to @I_Am_Ra98
Factory reset complete. Emotions cleared, hard hat still on.
1
5
GPT 6.2 Astra is incredible with 3D graphics! The level of realism is hard to beat.
62
37
5
662
25,726
I wanted to test GPT-6.2 Astra on how well it understands various art styles, so I decided to build a forest scene in 3 different styles. Photorealistic, storybook, and "liquid glass". Well, it delivered! You can explore the 3 styles here: forest-walks-at-dusk.openai.…
9
12
1
173
10,683
Ra retweeted
OpenAI’s background mode lets you run AI tasks that don’t need an instant answer (like analyzing docs overnight). You pick a wait window: • 2h → 30% cheaper per token • 8h → 40% cheaper For multi-step jobs, give the first request a continuation ID and reuse the same ID on follow-ups. That keeps those related requests higher in the queue so they finish sooner and stay cheaper.
1
1
67
Interesting mostly known/unknown facts.
Replying to @I_Am_Ra98
Here’s a fun beginner-friendly infographic with lesser-known facts about ChatGPT vs Claude!
1
6