Raphael Riad

@AurelienCa80656 @SplatLabs Did you use a 360 camera for the capture?

Here is a Gaussian Splatting capture of a construction site. I added project documents, AI-generated projections of the future building, and measurement tools inside a construction monitoring interface created by @SplatLabs .

@jun_song Sorry but opus for backend is undefeated. Deeseek flash for frontend is also great depending on your stack. Agentic and research deepseek flash also.

Best AI models now : Backend : GPT-5.5 Frontend : Kimi-K2.6 Agentic : MiMo-V2.5-Pro Best Price : Deepseek-V4-Flash Worst Price : Opus-4.7 Fast Research : Grok 4.3 Image : GPT-Image-2 Video : Seedance 2.0 (#1 for 3 months) Yes, Claude sucks.

@hiarun02 Because it cant. I’ve used over a billion tokens of deepseek v4 since it came out but opus is still undefeated. I think the gab is smaller then ever but still not 0

If DeepSeek V4 can do the same coding task for $5 why are people still paying $100 for Claude Code?

@OfficialNathanY Super cool project. Will definitely check it out.

What if your robot could understand any object you describe, just from a phone camera? RADIO-ViPE builds a 3D map from raw monocular video that you can query with natural language. (1/4)

@trikcode Surprisingly smart model. Has a lot of general knowledge even regarding some technical topics. Would not let it write any code for me tho

Has any developer tried this? I really want to know if this is good

Introducing O-Build. Something extraordinary. Something that changes everything about how we build. Think of your construction project. Now imagine having all your tools in one powerful platform - planning, management, tracking, everything. That's just the start.

Construction is about to change forever.

To build something is difficult, to build something new is almost impossible. But without innovation we fall slowly into obscurity. Excited to finally announce @OblixTech ! Construction is about to change forever.

@leecronin But you’re right it is a glorified search search engine at it’s core..

@leecronin You give a monkey a typewriter and enough time it’ll produce all of Shakespeare’s work. Same with an llm. What we need is a scalable validation system then maybe it’ll invent things

Chatbots cannot invent anything novel, they are probabilistic search engines.

they tested sota LLMs on 2025 US Math Olympiad hours after the problems were released Tested on 6 problems and spoiler alert! They all suck -> 5%

@skdh If we develop methods to validate an LLMs responses at scale it might even write you something new:)

@skdh You give a monkey a typewriter and enough time it’ll write you all of Shakespeare’s work

I've now received 4 "new physics theories" written "with the help of AI" in the past 2 weeks. This is going to be really bad.

I don’t think apple makes it in the AI age. Moving slow is fine when you’re adopting a new screen technology or a phone battery but AI follows different rules. The medal will go to whoever solves running good enough models on small modular devices. Essentially rabbit but real.

Just tried sesame ai and this is probably the infliction point for me. Thought we had more time..

LLM will never reach AGI, but they don’t need to. Applications built on top of them will be used to replace the majority of jobs that we have today. They will make us more efficient, powerful, and productive, but not AGI. At least this is what I hope 🤞

AGI is the wrong target. If you judge a fish by its ability to climb a tree it will live its life believing it’s stupid…

Here’s my humble opinion on why I don’t think LLMs can EVER reach AGI: