oxalpha is not frontier. I don't care what anyone says on a second run, Its still too dumb wrestle with vulkan swapchain fences and semaphores, execution barriers. Even tho Its been runnning non stop for 8 hours

🚨 RouteLLM API - An API that routes to your favorite models based on your prompt - create custom routers - OxAlpha serving tomorrow - Supports 150+ AI models - Excellent speed and prices - Handles caching - Works in Claude, Codex or anywhere You can pick and choose YOUR FAVORITES

Ox Alpha was doing 100T tokens/day, so many thought it was Gemini. Turns out it's GLM-5.3-Flash on Chinese chips. Jensen was right on Dwarkesh podcast: export controls may accelerate China’s NVIDIA-independent AI stack.

Ox Alpha was not a mystery model. It was a stress test of three things at once: - Can a Chinese lab ship a usable frontier model in stealth? - Can it absorb 20T+ tokens of global traffic for free? - Can it do that on domestic chips, not Nvidia? The answer, this week, was yes. China gave the world 20T+ tokens for free, claimed 100T/day capacity, then confirmed the inference ran on domestic chips. - You can argue about quality. - You cannot argue about the demonstration.

OX ALPHA IS NOW #1# ON OPENROUTER BY TOKEN USAGE, almost 2x the #2# model. Make a model cheap enough (free, in this case), and people will use it at massive scale without knowing who made it or what the official benchmarks are. As long as it's reasonably smart.

Ox Alpha built this very satisfying Powerwash Simulator no idea what changed but the outputs were not THAT good yesterday all assets generated from scratch: entire courtyard, the wall, the spray mechanics, the grime stripping off brick by brick could it be a new checkpoint? i'm not really sure.