32 total local models tested head to head
I ran 32 local models head to head on one fact-extraction corpus, 1,001 notes, paired bootstrap on every adjacent pair. Several weeks of compute time, all…
Latest
Non-Coding Harness Terminal UI
I am well versed in Cline, Qwen Code, Claude Code TUI experiences. I like the terminal. But their system prompts are refined for the developer experience.…
Unsloth’s Gemma 4 mmproj silently broke vision & audio on newer llama.cpp builds — anyone else hit this?
So I had been building ScreenMind, kinda like local ai desktop assistant that uses Gemma 4 for screen analysis, voice memo transcription, and meeting transcription —…
How come artificialanalysis.ai ranks Gemma4 above Qwen3.6 27b in SciCode
Just came across this coding benchmark: SciCode Artificialanalysis.ai reports a ranking which contradicts the feeling we've towards those models in real life coding. Is Gemma 4…
The death of SLMs?
I love to see these impressive models coming out that compete with the giants from companies like Z.ai, Moonshot, Alibaba, etc. A win for the open…
I built my own voice cloning model — looking for people to test it
My app - completely free, no signup Hey everyone, I’ve been working on a voice cloning model and finally got it to a point where other…
Knowledge vs. hallucination rate: what is your favorite model?
https://preview.redd.it/h7dwy09z2rhh1.png?width=1240&format=png&auto=webp&s=7285448ac2d9c895c9a11062557044fc97504fa3 I was evaluating which model to test for text summarization and rewriting in a scenario that also requires general world knowledge. Taking into account the…
They almost catched up on Frontier performance, so now catching up on prices
Users also report that the free version was significantly downgraded after the release of the new models this is very important for us when considering local…
Auto-fit vs tuned MoE offload: 564 → 1330 pp tok/s, unchanged decode (Qwen3.6-35B-A3B Q6 / RTX 3090)
TL;DR: On a Qwen3.6-35B-A3B Q6 setup sized for 64K context on a 24GB RTX 3090, spilling eight MoE expert layers to CPU freed enough VRAM to…
Final optimization: from ~10 tok/s to ~15 tok/s on DeepSeek-V4-Flash-0731 at 128K ctx – 1 RTX 3090
J'ai consacré beaucoup de temps à l'optimisation de DeepSeek-V4-Flash-0731 GGUF sur une seule RTX 3090. Mon exigence absolue pour chaque configuration était la suivante : Le…