delicious_apple

Joined 5 months ago

Karma

Submissions

Member since

May 21

Recent submissions

I don't agree with that conclusion. I'm looking at the benchmarks page for LFM2.5-24B-A2B, and it appears to get a worse score on SWE-Bench-Verified (by a lot) when compared to Qwen 3.6 35B-A3B. It only clearly beats Qwen 3.6 35B on AutomationBench, ToolSandbox, and LiveClaw.

At least for my use cases, Qwen 3.6 35B-A3B appears better than this new release. An MoE for Qwen 3.8 would be even better than that!

Comment · 2 months ago

I'm currently running it on an RTX 3090 (street price ~$1000 USD) with a long context and getting pretty good performance.

Prefill: ~1000 tok/s

Decode: 75-100 tok/s

It'll be far faster on a 5090, but I find the above performance to be acceptable. I've seen some claims that it even works OK on an AMD RX 7900XT (~$500USD)

Comment · 2 months ago

I am running it on a single RTX 3090 (24GB VRAM).

Some folks on Reddit are having the same experience: https://www.reddit.com/r/LocalLLaMA/comments/1vkm42m/muse_gl...

It uses an order of magnitude less VRAM at longer contexts which is a huge advantage over Qwen 3.6 27B

Comment · 5 months ago

You can always try openSUSE Slowroll (in beta), which is a rolling release that updates less frequently than Tumbleweed. It advertises better stability.

https://en.opensuse.org/Portal:Slowroll