Top
Best
New

Posted by erdaltoprak 5 hours ago

Qwen 3.8 27B(huggingface.co)
455 points | 320 commentspage 3
xlayn 4 hours ago|
The file "Just loads" on llama.cpp, the Unsloth https://huggingface.co/unsloth/Qwen3.8-27B-GGUF is an MTP file, I see mostly the same speed on pp and generation. There has to be something wrong with those benchmarks, I find extremely hard to believe a 27B model can work similar or exceed opus 4.6.
minimaltom 3 hours ago||
Worth distinguishing knowledge/task benchmarks from IF / agentic. It doesn't seem out of the question that you can have a small model thats generally good at instruction following and long-horizon agentic, as usually in those cases any requisite knowledge is in the context.

Most of the benchmark improvements afaict are in agentic and instruction following benchmarks.

cyanydeez 2 hours ago||
I think you've been drinking the "LLMs only improve by adding parameter counts" that SOTA labs are selling VCs to build data centers so they can keep eating through cash to their own benefits.

To the countrary, the reason Chinese models are excelling in the smaller area is because there's tons of fat in closed source models because of the crazy cash being thrown around.

There absolutely is space to improve intelligence and capabilities without lathering on more and more parameters.

selectively 1 hour ago||
[dead]
erdaltoprak 5 hours ago||
This is one of the most important model releases since most use cases don't need SOTA/Frontier

If you want Qwen3.8-27B Serving Configs for the DGX Spark vLLM NVFP4 and RTX 4090 llama.cpp GGUF I added the setups here https://x.com/ErdalToprak/status/2088299678085308761?s=20

kanemcgrath 2 hours ago||
I think I am going to buy a second rtx 3060, as 27B has been just outside of my range for to long, and this looks like the parameter count tipping point
apitman 1 hour ago|
Running it on 2x3060 now. Works pretty well but VRAM is tight. 4bit quants. 1x128k context, 8bit KV, MTP on.
kanemcgrath 1 hour ago||
whats the tok/s you get on that. I have heard a few claims of around 30-50 with mtp, but for how cheap the setup is I am surprised I don't hear more about 3060 stacks so I assume there has to be some catch.
apitman 1 hour ago||
I used GPT-5.6 Sol high to optimize it, and it claimed it was getting 50. I'm seeing ~40 on my goto smoketest: "Make me a vector add in CUDA".

Funny side note. It successfully one shot the program, but it wasn't able to run it because there literally wasn't enough VRAM left to allocate CUDA memory. Watching it try to debug that was fascinating. I'm pretty sure it would have killed the llama-server (and thus itself) if it hadn't been running in a separate container.

literoldolphin 2 hours ago||
Why is anyone even using video cards these days? You may as well be burning cash.

This is the perfect candidate for just splattering it on your nvme and then reading it off there and into memory. All of these run perfectly fine on simple m4 silicone:

https://github.com/drumih/turbo-fieldfare

https://github.com/leonickson1/Swiftlet

https://github.com/sqliteai/warp

awkwardpotato 2 hours ago|
Those are all for MoE models. And I prefer measuring my tokens in t/s instead of s/t
piyh 4 hours ago||
Qwen 3.6 is ~$2/m tok, 3.8 should be drop in replacement. Gemma 31B is $0.34/m tok. The price differential on these models is massive on openrouter.
SparkyMcUnicorn 2 hours ago||
Yeah, I would appreciate if someone could make sense of the pricing differences between these models. How can a provider run DSv4F at lower cost than a 27B dense or 35B A3B model?

Does it come down to utilization and/or specific model tricks and efficiencies (attention, kv cache, etc.)?

DeepInfra prices:

Qwen 3.6 27B: $0.32 in / $3.20 out

Gemma 3 27B: $0.08 in / $0.16 out

DeepSeek V4 Flash 0731: $0.08 in / $0.18 out

Qwen 3.6 35B A3B: $0.10 in / $0.95 out

https://openrouter.ai/qwen/qwen3.6-27b

https://openrouter.ai/google/gemma-3-27b-it

https://openrouter.ai/qwen/qwen3.6-35b-a3b

https://openrouter.ai/deepseek/deepseek-v4-flash-0731

mordae 2 hours ago||
DeepSeek V4 Flash is natively FP4 MoE with very compact KV cache. Say 8 GB/s. Qwen 27B is about 60 GB/s at full FP16 precision.
SparkyMcUnicorn 1 hour ago||
If active weight bandwidth is everything, then wouldn't Qwen 35b a3b (FP8) be roughly half the cost of dsv4f (FP4)?

dsv4f appears bigger and more memory/computationally expensive to run any way I look at it.

jjice 3 hours ago|||
Where do you see that? From what I can see on Open Router, Qwen 3.6 27B (the closest dense equivalent to Gemma 31) is $0.28/m. Am I missing something?

https://openrouter.ai/qwen/qwen3.6-27b

satvikpendem 2 hours ago|||
They're comparing Qwen 3.8 Max to Gemma 31B, fundamental mistake.
satvikpendem 2 hours ago||
Why are you comparing a 2.4 trillion Max model to a 31 billion model?
maherbeg 1 hour ago||
Does anyone have a https://tenstorrent.com/hardware/cards to try it on?
tosh 5 hours ago||
27b dense model at Opus 4.6 level

Opus at home

I hope there also will be a new ~10b variant

UncleOxidant 4 hours ago||
I'm hoping for a 3.8-122B MoE
yassa9 4 hours ago||
can you tell me ideas of usecases of 9 or 10B language models ? I cant find any usecases other than training a lora on them to give good bash commands for example
mring33621 4 hours ago|||
9B Qwen models are good and fast for local python coding tasks.
tosh 4 hours ago|||
they are all overlapping but:

categorization, information retrieval, semantic search, image description

also with the model as part of an agentic system with tool calling

(edit: it is quite impressive what a small model in a feedback loop can do)

fr2029 1 hour ago||
Will there be an A4B MoE?
syntaxing 2 hours ago||
Would I be surprised there’s bench maxing happening? Yes. But some users also use Q4 quantized and complain how dumb local models are.
chvid 5 hours ago|
These are massive improvements - and something you can actually run on a laptop.
More comments...