noclipepe

@Kamikaze6888 yeah definitely seeing a pattern here

@noclipepe Same one as before — cheaper Opus 5 outputting better 3D graphics than pricier Fable 5 is a solid result if it holds up 😄 Price-performance-wise this makes picking models less about "expensive = better" now.

In this round, Opus 5 completely destroyed Fable 5. Same prompt. Same test: "Build an ultra-realistic 3D flight simulator." The gap wasn't even close: Fable 5 ($2.82): 2x more expensive — broken aerodynamics, laggy controls, 2008-era graphics. Opus 5 ($1.40): 2x cheaper — realistic bank/pitch physics, smooth WebGL lighting, instant response. Opus 5 just completely reset the baseline for AI-generated 3D graphics. Which flight physics look better to you?

@baderslashxqt fair enough

@noclipepe Kimi win because he make the real tree

Claude Opus 5 just got crushed on speed by a model that cost under $1. Same prompt. Same React stack. Fresh terminals. Fable 5: $0.98, 310 seconds Kimi K3: $0.78, 1,512 seconds Opus 5: $2.21, 1,529 seconds Fable 5 was 5x faster than Opus 5 and Kimi K3. Kimi K3 wins on raw API cost ($0.78). But speed and price are not what you ship. Watch the results. Which model actually won?

Four AI configuration. An insane 3x gap The build times were nowhere close: Opus Max: 10:36 Kimi K3 Max + Pi: 16:56 Kimi K3 Max: 17:44 Opus Medium: 32:48 The timer has an obvious winner. The video might give you a different one. Which result would you actually ship?

Anthropic is still outside the Open Secure AI Alliance. OpenAI joined. Google joined. They joined NVIDIA and 36 founding partners in the Open Secure AI Alliance, launched today to share AI security models, agent harnesses, tools and research openly. The bet is simple: defenders should be able to inspect, adapt and run frontier systems locally, without one API provider setting the limits. The divide is now impossible to ignore: Open defense or gated capability. Does the last missing frontier lab join next?

The open-weight era just crossed a new line. Moonshot published the document explaining what is actually inside Kimi K3. 2.8T total parameters. 104B active per token. 16 of 896 experts selected. Native vision. 1M-token context. A model at this scale can now be inspected, modified and built on. Open weights are no longer the side story. They are becoming the main event. Read the report before your feed turns it into recycled takes

@knowix_eth And it was the cheapest one tho. kinda wild

@noclipepe all models cooked with this,but I like the kimi outcome more lol

@Saidi_Token that’s the leaning tree of Pisa actually

@noclipepe But why is Kimi's result so bad? I mean - just check it out how it built the tree? WTF with it?

This video makes the AI leaderboard look useless. Three frontier models got the same prompt. Each won a different game. Opus 5 built the richest world: varied terrain, wind-reactive banners, unit animations and hit reactions. GPT-5.6 Sol had the best UI, according to the author. Grok 4.5 finished fastest by a wide margin in this run. Same React stack. Fresh terminals. No external visual assets. There is no best model. Which result would you actually ship?

@Im_IrushiK Fair. But Fable was 5x faster. Doesnt that speed gap count for sth?

@noclipepe For me kimi won this one. Fable is shi😭

@Kestrelgdy Facts. One wins your time, the other wins your budget

@noclipepe 5x faster" and "wins on raw [quality]" pulling in opposite directions is the actual interesting part — most benchmark posts pick one axis and pretend the other doesn't exist.

@vin_diesel__fan More comparisons? Ofc I dont turn my back on family

@noclipepe is there a part two for this video somewhere

@MoorePavel26275 Yeah managed to cook up a complex site from one prompt. Crazy how it built it faster than Opus

@noclipepe sol is better

Is the new Claude Opus 5 worse than GPT-5.6 Sol? Same prompt. Same React stack. Two fresh terminals. Opus took 27 minutes. GPT took 11. Then the price meter made Opus look nearly 3x more expensive. The pricing was wrong. After correction, Opus came in around $13, while GPT still cost less. But did faster and cheaper produce the better site? Watch both builds and judge the results yourself. The exact prompt is revealed at the end. Which site would you ship?

@JensenHuang used his first post on X to explain why open models matter. Four months earlier, he explained NVIDIA’s strategy: Proprietary models create products. Open models let every industry, researcher, student, and country build on AI. If everything is closed, research and innovation stay behind the model owner. Open source is not charity for NVIDIA. It is how the AI market expands to the entire world. Here is his 84-second explanation.

Claude Opus 5 just dropped, claiming equal quality at half the price. The shift: Opus 5 isn't just cheap. It completely crushed Fable 5 at real-time physics while costing 2x less. Physics Prompts → Real-Time HTML Rendering → 2x Cost Cut → Full Physics Accuracy Four models faced the exact same benchmark: generate 3 self-contained HTML physics scenes. The results & costs: • Opus 5 ($1.40 / 55.9K tokens): The winner. Houses fly up the tornado, walls shatter on wrecking ball impact, and bridge rubble piles up realistically. • Fable 5 ($2.82 / 55.1K tokens): 2x the price for broken physics. The building collapsed on its own before the wrecking ball touched it. • GPT 5.6 ($0.31 / 20.1K tokens): Cheapest, but failed. The wrecking ball never reached the target. • Kimi K3 ($0.55 / 35.7K tokens): Same result as GPT—ended up with complete physics breakdown. If you build browser graphics or simulations, Opus 5 just reset the cost baseline.

Both @JensenHuang and @elonmusk are now aligned on Open-Weight AI. The shift: The closed API monopoly is collapsing. Hardware and compute titans are demanding Open Weights. Walled Gardens → Open Security → Sovereign Compute Why they're pushing for open models: • Security: Open weights let defenders audit code in real-time—secrecy creates systemic risk. • Sovereignty: Nations and enterprises must own their weights, not rent them via APIs. • Innovation: True scaling happens when the entire developer ecosystem can build freely. When the leaders of global AI compute tell you open weights are mandatory, the industry listens. Full letter attached below:

A Finance Professor from Marquette University just published a paper on Moonshot AI's Kimi K3. The shift: We’ve officially moved past simple code generation. Open-weight models are now autonomously engineering hardware. 2.8T MoE Architecture → Autonomous Chip Engineering → #1 Frontend Arena → $3/M Tokens Here are the 4 key takeaways from Professor Krause's 13-page paper: • Hardware Breakthrough: Kimi K3 autonomously designed a functional 4mm² silicon processor in a single 48-hour run and wrote its custom GPU compiler from scratch. • Frontier Parity: Reaches performance parity with proprietary models like GPT-5.6 Sol and Claude Fable 5. • Code Dominance: Ranked #1 on the arena.ai Frontend Code Arena with a 1679 score (beating Claude Fable 5's 1631). • Infrastructure Melt: Demand was so insane that Moonshot had to pause new subscriptions within 48 hours to save their GPUs. Open weights drop July 27th. If you think AI is still just writing basic web apps, you're missing the bigger picture. Read the full breakdown below.