
🐨
14.1K posts

🐨
@blclarke
Crypto & early tech enthusiast | AI agents, RWA, helping everyday degens | Bullish on $NULL $AVIOR 🚀 #DYOR #NFA #CT (will speak to retiree clients on outside)










@JensenHuang @nvidia Thanks for the free AI! Open models absolutely matter. But a “550B” badge without a clock, live tools or basic evidence discipline is less AliExpress-grade and more clearance-bin Temu edition. Frontier model. Discount execution path.


@wallstreetbets ok send me some thanks

@OpenAI’s ChatGPT web just “confessed” the exact execution-path split we kept running into while building VOOL: SOL 5.6 in Codex/ChatGPT via ChatGPT login ≠ SOL 5.6 via API key. Same model label. Different execution path—and, in our repeated real-world testing, materially different output quality. “Third-brother mode”** is real enough to measure. That means a model name alone tells you very little. The app, endpoint, harness, context handling and hidden limits can change the actual product you get. @AnthropicAI users: are you seeing the same split between Claude App and API/CLI? We haven’t noticed it ourselves so far. **Same model badge. Temu-grade execution path.











