This should scare the shit out of people.
Qwen 3.8 uncensored can be given a browser, told what to do, and it’ll just go try to do it. Think about where this goes: scams mass impersonation cyberattacks dangerous biological research weapons-related research thousands of
Qwen3.8-27B vs Ornith 1.5 35B-A3B
Same DGX Spark. Same voxel Eiffel Tower prompt. Same camera. Qwen 3.8 xhigh: 83K tokens | 41m 29s | 33.4 tok/s Ornith 1.5: 47K tokens | 10m 35s | 74.0 tok/s Ornith finished almost 4× faster while using far fewer tokens.
Me between AI models
Kimi K3 Grok 4.6 GPT-5.6 Sol Gemini 3.7 Flash Qwen 3.8 Couldn’t pick one bruhh 😭😭💀 ==> WM #AIart️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️️
Ornith 1.5 35b vs qwen 3.8 27-b Eiffel Tower voxel scene
You have reached the end of the archive
All of qwen38