DC

[DC @_davidcoles]

@Morrowmake Needs very little setup and auto self corrects itself, research is very thorough as well so far, can connect to Claude and many other accounts and lots more

[float @notesbyfloat]

A man cannot prompt the same agent twice, for it is has compacted and is not the same agent, and he is not the same man

[DC @_davidcoles]

@jakehemmerle @Morrowmake Ayeee let’s gooo!

[DC @_davidcoles]

@jakehemmerle @Morrowmake Neither do I man lol, just grabbed some measurements and a blender mcp and had Astra go to work lol, took a few partial test fit prints but I had it done in less than a day

[DC @_davidcoles]

@MrSage @Morrowmake Thanks! It has 352 gb of ddr4, (random mix of left over 32 and 64 gb dimms from work), 3945wx threadripper, 8 noctua fans, 2tb nvme, and 2 4tb hdd’s, and of course the 4 170hx’s

[DC @_davidcoles]

@jakehemmerle @Morrowmake Thanks! It’s a custom stl with 2 80mm arctic S8038 7k fans, noise isn’t too bad but it’s also not directly in my room I have it in a closet. If I get around to posting the stl I’ll be sure to share it here!

[DC @_davidcoles]

@Tech2Wild I use the full 5.3 at work and it’s so amazing, definitely recommend

[DC @_davidcoles]

@0xSero

[0xSero @0xSero]

Local AI X in 1 image.

[DC @_davidcoles]

@ClementPillette @0xSero What cards are those holy

[DC @_davidcoles]

@sugarflows @itsjustmarky They will get super hot, tried putting 4 next to each other and the middle cards won’t make it

[akhila_datta @Akhila_988]

Introducing Jev-Omni, the first multimodal system one model. (Other OSS versions miss atleast a modality)

Supports all modalities: text, images, audio and video !

On Par with Jev on Typed-benchmarks.
Scaled -> 30k examples on 8xH200 (data mix matters a lot)
< 100ms on 1 H100

https://t.co/ACieaM4hTi
More work is coming, so follow along !

[Charly Wargnier ☄️ @DataChaz]

anyone wanting to fine-tune an LLM on their own data after reading this:

[Mia @MiaAI_lab]

ZCode is now open-source! 🔥🔥

@Zai_org did it 👏🏻

This is the way.

https://t.co/owyE5Uwrx1

[DC @_davidcoles]

Been In the process of training something very similar

[Archer @4rcherhume]

Yes we are a healthcare company. Yes we just released the best open-weight alternative to Jev, Solomon 27b.
- It roughly matches the intelligence of its base-model, Qwen3.8 27b
- Natively Multimodal
- 265k Context Window
- Adds multi-choice tagging, evidence pointer spans, and more.

[DC @_davidcoles]

@stefanopineda Can’t escape the loop

[Archer @4rcherhume]

All the grifters are completely wrong about Jev’s architecture so I decided I’d release an open-weight version. BUT training takes time, so while we all wait I decided I’d drop the sauce.

https://t.co/TK9FIiUxAJ

[DC @_davidcoles]

Glm 5.3 local on 8 mi300x’s, what a time to be alive