I like to tinker. 4 decades in tech and really want to play with AI local LLMs from ~30B to ~122B. I also want to play with comfyui templates. I don’t need this for work. This is a lab for me to tinker with. But with ramageddon I do not have the money to link 2x DGX Sparks and I would like to upgrade my M2 Max 32Gb (which does 35B models on Ollama well).
Basically I am looking for a small form factor AI do it all hardware. Primary use case is Vibe coding but I do want to play with everything. Again it doesnt have to be blindingly fast. It needs to be usable. I keep going back and forth.. wait for rtx spark and roll the dice or just get an Asus Ascent and wait for M7 next year.
Thoughts?
source & further reading
forum.level1techs.com — original article
Dual Sparks in nvfp4 vs 4x RTX Pro 6000 with native DeepSeek V4 0731 -- Quants and Speed
Which do you think is better: a large local language model or a cloud-based API?
Budget GPUs or use spare 4090 for local model? Use case included.