Sudo su
Verified account This account is verified. Learn more@sudoingX
GPU/local LLM. more RAM and OSS... everywhere
1.2K Following 40K Followers 13K Posts
Latest posts 5
i'm waiting for qwen 4 27b dense like a man waiting for his long distance girlfriend to land.
some people are preparing dinner. some people are preparing for a date.View quoted post on X
i am preparing 24GB of VRAM for qwen 4 27B dense model.
we all have different priorities anon.
some people are preparing dinner. some people are preparing for a date.
i am preparing 24GB of VRAM for qwen 4 27B dense model.
we all have different priorities anon.
i am preparing 24GB of VRAM for qwen 4 27B dense model.
we all have different priorities anon.
BREAKING: sources close to my heart confirm qwen 4 27b drops november 8, 11pmView quoted post on X
okay! this one is interesting if you run 2x dgx sparks
ant ling open sourced ling 3.0-flash sante, their health model, and the official fp8 is 128.4 gb, too big for one spark and it fits across two
> 124b total, 5.1b active
> 256k context, mit license
i boxed the 4 panels where it beats every open model on their chart, kimi k3 at 2.8t included, and two of the three closed ones
> healthbench pro 54.3 vs opus 4.8 49.4
> diagnosisarena 79.2 vs kimi k3 78.4
> hle w/ tools 57.1 vs gpt-5.6 sol 56.6
> gpt-5.6 sol ahead on 3, opus on hle
those are their numbers, the weights are already pulled on both my sparks and i'm testing it now
ant ling open sourced ling 3.0-flash sante, their health model, and the official fp8 is 128.4 gb, too big for one spark and it fits across two
> 124b total, 5.1b active
> 256k context, mit license
i boxed the 4 panels where it beats every open model on their chart, kimi k3 at 2.8t included, and two of the three closed ones
> healthbench pro 54.3 vs opus 4.8 49.4
> diagnosisarena 79.2 vs kimi k3 78.4
> hle w/ tools 57.1 vs gpt-5.6 sol 56.6
> gpt-5.6 sol ahead on 3, opus on hle
those are their numbers, the weights are already pulled on both my sparks and i'm testing it now
Today, we’re officially open-sourcing Ling-3.0-flash-Sante.
Sante is a healthcare-enhanced MoE model with 124B total parameters and only 5.1B active parameters per token, delivering leading performance among similarly sized models on professional healthcare tasks and complex medical reasoning.
Beyond answering medical questions, Sante is built for real-world healthcare research and workflows, with strengthened capabilities in deep research, multi-agent collaboration, and tool use.
The model weights are now available — we welcome researchers, developers, and healthcare practitioners to explore Sante and build with it.
More below ↓View quoted post on X
DEVELOPING: my 3090 has started pre-warming fans at 30%. 24gb cleared. this is not a drill
BREAKING: sources close to my heart confirm qwen 4 27b drops november 8, 11pmView quoted post on X