points by rglover 4 days ago | 0 comments
"Cheap" is up for debate, but the models themselves are great. Kimi K3 (and now 3.1 Pro) was the first open model I played with and didn't think to switch back to a frontier model. Now it's my daily driver and I have zero complaints.
The frontier labs are in a tough spot. It's hard to justify their cost when you can get something that's easily as good for cheaper (and anecdotally, more reliable/accessible). I do HITL so my perspective will be different than a fully-agentic setup, but as soon as I started using open stuff regularly, I realized that any moat the US labs had built is drying up.
[–] vorticalbox 4 days ago
[–] dofm 3 days ago
Though if you have the room for it, Muse Glimmer is better in practice on low reasoning strength, and while its token generation is slow, it actually gets to answers pretty quickly.
[–] vorticalbox 3 days ago
[–] dofm 3 days ago
"There are dozens of us!"
The thing you need to know about Muse is that you control its reasoning/thinking depth in the system prompt (low/medium/high/xhigh):
… is pretty responsive and I've found its time to first answer to be surprisingly good.It is broadly comparable to Qwen 3.8 27B on "low" thinking, though it is weaker on very long context, one-shot stuff, but I dunno… I just like it more as a developer companion.
(I'm using Unsloth's version)
[–] vorticalbox 21 hours ago