Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Qwen3.6-35B-A3B: Agentic coding power, now open to all

qwen.ai

Post reply on HN
Print view View original

491–500 of 563 posts

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#491

Post by wolfhumble »

kylehotchkiss wrote: post #252
zozbot234 wrote: post #40

Earlier quoted context omitted.

This is just one model in the Qwen 3.6 series. They will most likely release the other small sizes (not much sense in keeping them proprietary) and perhaps their 122A10B size also, but the flagship 397A17B size seems to have been excluded.

How many people/hackernews can run a 397b param model at home? Probably like 20-30.

This is like saying that Open source is not important because I don't have a machine to run it on right now. Of course it is important. We don't have any state of the art Language models that are open source, but some are still Open Weight. Better than nothing, and the only way to secure some type of privacy and control over own AI use. It is my goal to run these large models locally eventually; if they all go away that is not even a possibility. . .

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#492

Post by Ladioss »

txtsd wrote: post #306
Ladioss wrote: post #175

Earlier quoted context omitted.

More like `ollama launch claude --model qwen3.6:latest` Also you need to check your context size, Ollama default to 4K if <24 Gb of VRAM and you need 64K minimum if you want claude to be able to at least lift a finger.

I only have 16GB VRAM, and my system uses ~4GB from that. What are my options? I got this one: `Qwen3.6-35B-A3B-UD-IQ2_XXS.gguf`

My system has 16 Gb VRAM / 32 Gb RAM, and ollama runs qwen3.6:latest at decent speed just fine. The 35b model is a moe, so I guess the whole model is offloaded.

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#493

Post by redman25 »

zozbot234 wrote: post #319
deaux wrote: post #291

Earlier quoted context omitted.

> If you just do a tiny amount of tok/day and can wait for the answer to be computed overnight or so But they can't? The usage pattern is the polar opposite. Most people running these models locally just ask a few questions to it throughout the day. They want the answers now, or at least within a minute.

If you want the answer right now, that alone ups your compute needs to the point where you're probably better off just using a free hosted-AI service. Unless the prompt is trivial enough that it can be answered quickly by a tiny local model.

A strix halo machine or MAC will run at less than 20watts idle. You could leave it running.

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#495

Post by petu »

johanvts wrote: post #469
bertili wrote: post #285

Earlier quoted context omitted.

It's fascinating that a $999 Mac Mini (M4 32GB) with almost similar wattage as a human brain gets us this far.

Interesting thought, I looked it up out of curiosity and fund 155w max (but realistically more like 80w sustained) for the mac under load, and just around 20watts for the brain, surprisingly almost constant whether “under load” or not.

> 155w max (but realistically more like 80w sustained)

155W PSU seems to be unified with M4 Pro model, plus there's reserve for peripherals (~55W for 5 USB/Thunderbolt ports).

Apple lists 65W for base M4 Mac itself: https://support.apple.com/en-am/103253

Notebookcheck found same number: https://www.notebookcheck.net/Apple-Mac-Mini-M4-review-Small...

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#496

Post by reissbaker »

Der_Einzige wrote: post #172
ekianjo wrote: post #124

Earlier quoted context omitted.

yeah the 27B feels like something completely different. If you use it on long context tasks it performs WAY better than 35b-a3b

I've been telling analysts/investors for a long time that dense architectures aren't "worse" than sparse MoEs and to continue to anticipate the see-saw of releases on those two sub-architectures. Glad to continuously be vindicated on this one. For those who don't believe me. Go take a look at the logprobs of a MoE model and a dense model and let me know if you can notice anything. Researchers sure did.

Dense is (much) worse in terms of training budget. At inference time, dense is somewhat more intelligent per bit of VRAM, but much slower, so for a given compute budget it's still usually worse in terms of intelligence-per-dollar even ignoring training cost. If you're willing to spend more you're typically better off training and running a larger sparse model rather than training and running a dense one.

Dense is nice for local model users because they only need to serve a single user and VRAM is expensive. For the people training and serving the models, though, dense is really tough to justify. You'll see small dense models released to capitalize on marketing hype from local model fans but that's about it. No one will ever train another big dense model: Llama 3.1 405B was the last of its kind.

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#497

Post by survirtual »

jubilanti wrote: post #302
simonw wrote: post #270

I've been running this on my laptop with the Unsloth 20.9GB GGUF in LM Studio: https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/blob/mai... It drew a better pelican riding a bicycle than Opus 4.7 did! https://simonwillison.net/2026/Apr/16/qwen-beats-opus/

I wonder when pelican riding a bicycle will be useless as an evaluation task. The point was that it was something weird nobody had ever really thought about before, not in the benchmarks or even something a team would run internally. But now I'd bet internally this is one of the new Shirley Cards.

I use this metric now, and I suggest you change it per your imagination:

"Make a single-page HTML file using threejs from a CDN. Render a scene of a flying dinosaur orbiting a planet. There are clouds with thunder and lightning, and the background is a beautiful starscape with twinkling stars and a colorful nebula"

This allows me to evaluate several factors across models. It is novel and creative. I generally run it multiple times, though now that I have shared it here, I will come up with new scenes personally to evaluate.

I also consider how well it one shots, errors generated, response to errors being corrected, and velocity of iteration to improvement.

Generally speaking, Claude Sonnet has done the best, Qwen3.5 122B does second, and I have nice results from Qwen3.5 35B.

ChatGPT does not do well. It can complete the task without errors but the creativity is atrocious.

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#498

Post by egorfine »

Patrick_Devine wrote: post #276
Ladioss wrote: post #175

Earlier quoted context omitted.

More like `ollama launch claude --model qwen3.6:latest` Also you need to check your context size, Ollama default to 4K if <24 Gb of VRAM and you need 64K minimum if you want claude to be able to at least lift a finger.

If you're on a Mac, use the MLX backend versions which are considerably faster than the GGML based versions (including llama.cpp) and you don't need to fiddle with the context size. The models are `qwen3.6:35b-a3b-nvfp4`, `qwen3.6:35b-a3b-mxfp8`, and `qwen3.6:35b-a3b-mlx-bf16`.

I was comparing various models at M5 Pro 48GB RAM MLX vs GGUF and found that MLX models have a higher time to first token (sometimes by an order of magnitude) while tokens/sec and memory usage is same as GGUF.

Gemma 3 27B q4:

* MLX: 16.7 t/s, 1220ms ttft

* GGUF: 16.4 t/s, 760ms ttft

Gemma 4 31B q8:

* MLX: 8.3 t/s, 25000ms ttft

* GGUF: 8.4 t/s, 1140ms ttft

Gemma 4 A4B q8:

* MLX: 52 t/s, 1790ms ttft

* GGUF: 51 t/s, 380ms ttft

All comparisons done in LM Studio, all versions of everything are the latest.

Re: Qwen3.6-35B-A3B: Agentic coding power, now open to all

#500

Post by logicallee »

simonw wrote: post #270

I've been running this on my laptop with the Unsloth 20.9GB GGUF in LM Studio: https://huggingface.co/unsloth/Qwen3.6-35B-A3B-GGUF/blob/mai... It drew a better pelican riding a bicycle than Opus 4.7 did! https://simonwillison.net/2026/Apr/16/qwen-beats-opus/

what kind of specs does your laptop have? do you know how many tokens/second you get on it?
Post reply on HN