BrainOS

How can we debug the operating system in our own brains? This is the most important question facing humanity.

[BrainOS @BadBrainCode]

@TheAhmadOsman I’ve been having a blast with qwen3-8b-int4-ov running via openVINO in my small 16gb Intel ARC box. It’s snappy and fun.

[On yedinci Mesele @sekizaltmis_]

@witcheer 150-200+ tok/s @SpaceTimeViking

https://t.co/w7nps0Ophk

[BrainOS @BadBrainCode]

@0xSero I’m in for free.

[BrainOS @BadBrainCode]

Running Gemma-4-26B-A4B-NVFP4 on my DGX Spark GB10 via vLLM.
Results: ~30 tok/s single-stream, 53 tok/s at c=2.
For context: Nemotron-3-Super 120B on the same hardware does 14 tok/s.
The MoE architecture is the reason — only 3.8B of 26B parameters activate per token step. Memory bandwidth stops mattering when you’re barely touching the weights.
18 GB weights. 2-min cold start. Full 128k context, parallel tool calls. Native image, audio, and video support. Should I try FP8 + MTP?

[BrainOS @BadBrainCode]

@jessegenet Will winning the bit manipulation race ensure we win the atom manipulation race?

[BrainOS @BadBrainCode]

@0xSero I had a shocking Orwellian experience yesterday where Claude got righteous with me. It scared the crap out of me. No doubt it would have contacted my boss if it could’ve. I’m unplugging from Anthropic. Open source will save humanity. No doubt in my mind.

[Teknium 🪽 @Teknium]

@uzairansar PSA: This is not an official app by us - it would be nice uzi if we made this more clear in the name or something

[BrainOS @BadBrainCode]

@cursor_ai Composer 2.5 is a Rocketship in grok CLI. Fired it up for the first time and it is blazing.

[BrainOS @BadBrainCode]

@petergyang @WisprFlow Huge upgrade. Auto find was awful.

[BrainOS @BadBrainCode]

@garrytan Every session should be summarized at 3:00am and saved as a local markdown. I have claude trained on that well.

[BrainOS @BadBrainCode]

@0xSero I’m assuming you mean you can remote access a codex or claude session running on a local model. It’s really cool. I have had issues copying text on mobile though.

[BrainOS @BadBrainCode]

@mr_r0b0t Have you been able load two models with no degradation of performance?

[BrainOS @BadBrainCode]

@mr_r0b0t Thank u this is yuge.

[BrainOS @BadBrainCode]

@Pontifex This won’t age well.

[BrainOS @BadBrainCode]

@TeksEdge Holy Fuck. They won’t be able to produce those fast enough. Get ready for the best selling product launch in history.

[BrainOS @BadBrainCode]

@SlimTradeyBaby @0xSero Have you achieved concurrent models on a DGX? I hit a hard roadblock.

[BrainOS @BadBrainCode]

This is a much watch for anyone considering local inference. Excellent presentation. Great job.

[0xSero @0xSero]

I yapped about LLM Compression for 40 minutes, how much misinformation did i spread this time (,:

[BrainOS @BadBrainCode]

@mr_r0b0t I have Nemo 120 working with claude code harness. Slow but functional. 14tps decode.

[BrainOS @BadBrainCode]

@mr_r0b0t If you can validate agentic capabilities I won’t be able to resist buying a second. It’s only going to get better from here.

[BrainOS @BadBrainCode]

@0xSero Fascinating. 16Bit numbers.