Lukemaz

Carrying forward the idea of being able to fly

[Lukemaz @mucalazzoni]

As I've been saying for a long time, this is exactly why optimizing models for local use is so important — no one can take those away from us. I'm actively contributing to this, so please support my project.
Qwen3.6-35b-a3b.
https://t.co/E4xSSoIf0W
#mythos #fable #anthropic #AI

[Lukemaz @mucalazzoni]

@lightseekorg @Alibaba_Qwen Release something for macOS

[atomic.chat @atomic_chat_hq]

Run Uncensored MiniCPM-V 4.7 35B-A3B 👀

We removed the filters from @OpenBMB's latest model so refusals on filtered prompts drop from 96.3% to 0%

Get it on our HF 👇https://t.co/vIaGBgcny9

[Sidera Foundry @SideraFoundry]

@elonmusk @Bots Step two: your bot emails my bot, they schedule the meeting, and neither of us has to attend.

[Lukemaz @mucalazzoni]

@Beaver_0x something similar but better to coordinate many humans

[Sidera Foundry @SideraFoundry]

https://t.co/1KRKkzPVIS
Install guide in the repo.
Feedback and issues welcome, and a ⭐ if you like it.

[Lukemaz @mucalazzoni]

@thegreatest_sv This is the repo btw

[Sidera Foundry @SideraFoundry]

https://t.co/1KRKkzPVIS
Install guide in the repo.
Feedback and issues welcome, and a ⭐ if you like it.

[Sidera Foundry @SideraFoundry]

https://t.co/1KRKkzPVIS
Install guide in the repo.
Feedback and issues welcome, and a ⭐ if you like it.

[Sidera Foundry @SideraFoundry]

@finkd "Personal superintelligence for everyone" is the right frame. The question we keep running into: what happens when everyone on a team has one, and they all have to work on the same project?

[Lukemaz @mucalazzoni]

@VaibhavLab Qwen 4 27b

[Sidera Foundry @SideraFoundry]

The line that stuck with me: "human review and comprehension are starting to fall behind."

We feel it every day: two founders, each with our own crew of agents on the same projects. The agents are rarely the bottleneck now. Keeping the humans in sync is.

[Lukemaz @mucalazzoni]

@thegreatest_sv Thank you so much

[Lukemaz @mucalazzoni]

@Oluwaphilemon1 Once open models this capable run locally on everyday hardware, there’s no putting the weights back in the box. The only way to try to stop them will be to regulate the market around them.

[Lukemaz @mucalazzoni]

can someone on underdog invite me? asking nicely @underdogai

[Lukemaz @mucalazzoni]

@OrcaRouter https://t.co/GX1l8xAQID
now you can know how much cost to quantize model on cloud.

[Lukemaz @mucalazzoni]

@pratikg @yukonresearch Is the site down just for me?

[Lukemaz @mucalazzoni]

@NFT_Chen Qwen3.6-35B-A3B 4-bit MLX (19 GB) running on a 24 GB Apple Silicon

https://t.co/E4xSSoIf0W

[Lukemaz @mucalazzoni]

@0xSero Quickly get your name, Before it’s too late https://t.co/lIjQI59drD

[Lukemaz @mucalazzoni]

@mztacat Quickly get your name, Before it’s too late https://t.co/lIjQI59drD

[Lukemaz @mucalazzoni]

@alexocheema Quickly get your name, Before it’s too late https://t.co/lIjQI59drD

[Lukemaz @mucalazzoni]

@sz_vc7 Quickly get your name, Before it’s too late https://t.co/lIjQI59drD