Does anyone have an idea how this might perform on a DGX Spark at longer contexts? I've been trying to investigate their performance with these medium-sized MoE models, but I'm seeing a lot of incomplete and conflicting information. The 273 GB/s bandwidth looks awfully bad on paper...
Qwen3.8-Flash-Next
61–70 of 246 posts
Re: Qwen3.8-Flash-Next
#62Adding to my homelab stack, hopefully it doesn't overthink like the little model. Actually, hoping it thinks a bit less. Wait actually I'm really praying it reasons a bit more directly. But wait, I'm really sure that it must be a bit better.
You’re absolutely right to be hopeful. Three honest possibilities, and I’ll be straight with you about each: 1. It overthinks — Just like the previous iteration. High confidence. 2. It doesn’t overthink — Improvement from the last model for your use case. Regression for others. 3. It sometimes overthinks — Best case all around. A feature, not an impairment. One final thing worth mentioning: (I made myself irrationall…
This is glitch art for text, I love it
Re: Qwen3.8-Flash-Next
#63Adding to my homelab stack, hopefully it doesn't overthink like the little model. Actually, hoping it thinks a bit less. Wait actually I'm really praying it reasons a bit more directly. But wait, I'm really sure that it must be a bit better.
My stack is basically deer-flow with Qwen3.5-122B-A10B; this hopefully will be a speed and intelligence improvement. Running deer-flow overnight on any research topic or verify clear scoped programming issue is really neat. Also, heating my home during the winter is nice. Oh, also, I use llamacpp with --reasoning-budget; very simple way to move on.
Yeah 122B is the sweet spot for me as well. Even deepseek flash overthinks on stuff way too much. I think they fully rely on large reasoning turns to achieve better quality. The result of course means we wait a long time to get results even with high throughput as a lot of tokens are wasted.
Re: Qwen3.8-Flash-Next
#64Also announced GLM-5.3-Flash: https://news.ycombinator.com/item?id=49449507
Re: Qwen3.8-Flash-Next
#65FYI: nothing seems to be able to run this (easily) yet. llama.cpp, vllm etc I couldn't get working because of no support in the mainline version.
They are giving pointers to how to run it now using for example https://recipes.vllm.ai/Qwen/Qwen3.8-Flash-Next (and an especially provided vllm release).
ah I don't think that page was up when I checked, it 404ed!
Re: Qwen3.8-Flash-Next
#66Aaaaaaaaaaaaaand dario meltdown on twitter in 3, 2, 1...
Re: Qwen3.8-Flash-Next
#67> Qwen3.8-Flash-Next features a 125B-parameter main model, supplemented by an additional 51B N-gram embeddings, with 6B parameters activated per token. Didn’t see this mentioned yet. I wonder what this means for the effective size. It’s evidently ~176B paramètres, but how does that get quantized. A 4-bit quant under 100GB seems unlikely, I’m suspecting this won’t run in 128GB unified memory In principle I like the id…
Gonna have to wait a few days to see what the wizards of the HF community come up with…
Re: Qwen3.8-Flash-Next
#68> Qwen3.8-Flash-Next features a 125B-parameter main model, supplemented by an additional 51B N-gram embeddings, with 6B parameters activated per token. Didn’t see this mentioned yet. I wonder what this means for the effective size. It’s evidently ~176B paramètres, but how does that get quantized. A 4-bit quant under 100GB seems unlikely, I’m suspecting this won’t run in 128GB unified memory In principle I like the id…
The N-gram parameters can be fetched from SSD, with maybe the hottest ones staying in memory.
Re: Qwen3.8-Flash-Next
#69Didn't expect it to beat 3.8 27B so cleanly. Opus 4.6 Max self-hosted at 30 tok/s on a 5k Macbook in Aug 2026. The LLM timelines are crazy.
My AMD strix halo box (haven’t benchmarked yet) should also run it reasonably well. It was $1400 at launch, and is $4K now.
Your mac is < $2K in Biden-era dollars. Presumably the economy will eventually recover; maybe in one Moore’s law doubling if the midterms go outrageously well. That’ll be two doublings since the halo launched. I’d expect this model to run on a sub $1K box by then. $2K ought to get you a 512b parameter model at that point. If we have to wait out the rest of the term, the cost cliff will be even more pronounced when it hits.
Re: Qwen3.8-Flash-Next
#70Didn't expect it to beat 3.8 27B so cleanly. Opus 4.6 Max self-hosted at 30 tok/s on a 5k Macbook in Aug 2026. The LLM timelines are crazy.
I don't like these comparisons. Sure it is impressive, but it does not have a world knowledge of larger models. It has most of theirs intelligence.
In a lot of spaces, this is actually preferable.
Ex - nodejs natively supports a huge set of typescript with built-in type stripping these days. But ask most hosted models to build a typescript project and they default to a heavy compile step, or a tool like tsx, ts-node, etc.
Models with lots of "world knowledge" have a good chunk of that knowledge go stale, and there's no real way to refresh it without training a new model.
Another classic example of this back in the day was to ask who the president of the US was, and watch different models happily give different answers based on the date they were trained.
---
Personally, I'm really interested to see if we're headed towards a spot where the model is entirely distinct from the knowledge store.
We're vaguely there with the ability for models to go search the web, but I think the reliability of that path is going to continue declining (more and more spam content, less and less genuine value).
I kinda want a paradigm where I can pick and engine and a knowledge bank, and combine them as I please.
Ex - if I'm doing gardening, I can pick "gardening for models (version 32)" as my knowledge store.
If I'm doing auto-repair... "cars for dummies (version 3)". etc...