Pinned
Okay, my co-founders might not win any Oscars.
Yet we created OORB Studio, an agentic platform for anyone to build, simulate, and deploy robots from the cloud (with project-wide context coming soon!)
try the public beta now → oorb.io
Shade
@shade_engine
Robotics and Deep learning, Local compute | coding oorb.io
Hannibal ad portas.
If you don't own your intelligence, you own nothing
bay area
Joined July 2018
- good morning sunshine 🌞Fellow Local AI enjoyers, brace yourselves! Looks like Qwen 3.8 35B-A3B is coming! Several repositories including references to the model have begun appearing all over Github 😁 @Alibaba_Qwen tease!
- finally got it running ( one gpu only sadly cuz the PSU I got doesn't support two ( the strix had 3 x 8 pins ) the next one comes monday ) > Q4_K_M unsloth gguf fits in ~17 gb of VRAM > running on a llama.cpp server + opencode for some fast tasks > 128k context window willwill not leave the office today till qwen is ripping through the gpus
- will not leave the office today till qwen is ripping through the gpusWe promised open weights for Qwen3.8. Now, time to meet them! 🎉 ⚡ Qwen3.8-27B: - A native multimodal dense model. With just 27B parameters, it outperforms Qwen3.7-Plus overall and shines in real-world coding & office workflows. - 262K native context, easily extendable to 1M
