antirez/ds4

Dwarfstar already supports this, curious how it compares, but I use the q4 quant daily and it works really well. https://github.com/antirez/ds4/blob/main/docs/MODELS.md#qwen38-flash-next

It is pretty nifty. I spend some time over last weekend implementing fused TQ to allow for 1m context lengths on a 128 gb MacBook M5 Max when using Qwen 3.8 flash next ( https://github.com/antirez/ds4/pull/1115 if you are interested). If I get bored I might port over the Metal kernels from oMLX -- the speed increase they have for the v0.7.0 release is amazeballs.

Primarily used ds4[0] by antirez [0] https://github.com/antirez/ds4

Counterpoint, I'm just a happy user of this local LLM [1] from antirez (of Redis fame). [1] https://github.com/antirez/ds4

I have two 128gb Strix Halos and have been extremely excited about Antirez's (Redis author) work on DS4, especially with 4bit quant using two machines: https://github.com/antirez/ds4 Right now the speed isn't good for GLM 5.2, Deepseek V4 Flash speed is okay for me (actually reading the output) and quite usable. See kyuz0's great recent video here: https://www.youtube.com/watch?v=PkKXm_mKCCM With a bit more speed and model improvements, local AI becomes a reasonable practical thing! The biggest problem is all the tech companies making consumer hardware completely unaffordable, and I don't think this is accidental. Look at Micron's profits and share price lately... I got my Strix machines …

I started with antirez' DwarfStar[1] on one spark and that (~11-14tok/s generation, ~300-400 tok/s prompt processing) was enough of a taste for me to jump into 2 sparks, running the native quant of DSv4 Flash. Now at 40-50tok/s generation and ~2000 tok/s prefill with a model that I've seen reason through race conditions and be able to trivially pull off any straight-forward coding task, and remain coherent at 500k context. With a preview checkpoint of the weights! I'm excited for the future of local LLMs. There is some buy-in but apparently not an extreme amount to get access to models that can stand in the for the giants on all but the most challenging and/or hands-off coding tasks. [1]:…

indeed, there's even a (pretty solid) custom server just for DS4 https://github.com/antirez/ds4 -- works very well on high-RAM Macs

ds4: prefill: 150.18 t/s, generation: 48.47 t/s together with https://github.com/antirez/ds4 + RTX Pro 6000, local provider :) I guess you could rent one too, get similar speed. https://gist.github.com/embedding-shapes/eceefa1b2699dca448f9af22e995cdeb The weights I've tried is the "DeepSeek-V4-Flash-IQ2XXS-w2Q2K-AProjQ8-SExpQ8-OutQ8-chat-v2-imatrix" ones which fits just about within 96GB VRAM. With some tuning, I've managed to get it to get up to ~60 t/s, I'm sure there is other things to do there too :)

> Trying to run them on a unified memory Mac > but still not quite in the realm of Sonnet or DeepSeek 4 Flash these are not mutually exclusive anymore. DS4 has set the bar for me these days. https://github.com/antirez/ds4