Hackernews posts about Qwen3

1. Qwen 3.6 27B is the sweet spot for local development (quesma.com)
   1192 points by stared 9 days ago | 758 comments
2. RTX 5080 and RTX 3090 Setup: 80 Tok/s on Qwen 3.6 27B Q8 (imil.net)
   292 points by iMil 25 days ago | 108 comments
3. Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions (www.teachmecoolstuff.com)
   215 points by dev-experiments 17 days ago | 49 comments
4. Rio de Janeiro's city government model Rio3.5 beats Qwen3.7 in recent benchmarks (twitter.com)
   142 points by lucasfcosta 24 days ago | 45 comments
5. Two Qwen3 models on one DGX Spark: the residency math (www.devashish.me)
   95 points by devashish86 20 days ago | 49 comments
6. Running Claude Code Offline on an M3 Pro with Qwen3.6 (har-ki.github.io)
   18 points by har-ki 27 days ago | 10 comments
7. ThinkingCap-Qwen3.6-27B: Qwen3.6 capabilities with 50% fewer thinking tokens (huggingface.co)
   5 points by victormustar 2 days ago | discuss
8. Show HN: FlashQwen – A from-scratch CUDA inference engine for Qwen3 (github.com)
   5 points by langtang1996 23 days ago | discuss
9. Running Qwen 3.6 Locally on a Mac Mini M4 with 16GB RAM (maloyan.xyz)
   4 points by mpweiher 5 days ago | discuss
10. Running Qwen 3.6 Locally on a Mac Mini M4 with 16GB RAM (maloyan.xyz)
    4 points by mpweiher 6 days ago | discuss
11. Qwen 3.6 93B with MTP on 2×RTX 3090 NVLink=187 tokens/SEC,LLM lost bleat-a-thon (github.com)
    4 points by devilfileprong 24 days ago | discuss
12. Norm-preserving abliteration on Qwen3.6-35B-A3B with 0% refusal (www.reddit.com)
    3 points by satvikpendem 5 days ago | discuss
13. Local LLMs on a Ryzen 8700G iGPU: 13-15 tok/s on gemma4, 9-12 on qwen3.6 (loomcycle.dev)
    3 points by denn-gubsky 8 days ago | discuss
14. Show HN: Qwen3.6 27B on a DGX Spark writing code that ships (www.devashish.me)
    2 points by devashish86 1 day ago | 1 comments
15. Qwen3.5 2B burns all the output tokens while thinking
    2 points by adithyaharish 9 days ago | discuss
16. Qwen3.5 WebGPU (huggingface.co)
    2 points by fagnerbrack 12 days ago | discuss
17. Indian Address Parser – Qwen3-0.6B LoRA (huggingface.co)
    1 points by gagan2020 7 days ago | 1 comments
18. Show HN: Learn from 30 historical figures, open source, nonprofit, self-hosted (github.com)
    49 points by micstradev 29 days ago | 24 comments
19. Show HN: An unmetered LLM API–$6/month, no token tracking, no limits (yolo-auto.com)
    12 points by yolo-auto 3 days ago | 13 comments
20. Show HN: How clanker are you? A reverse Turing test (howclankerareyou.com)
    11 points by niklio 5 days ago | 11 comments
21. Show HN: I am running 3 coding agents non-stop over the last 3 days. Here is how
    10 points by sermakarevich 25 days ago | 3 comments
22. Ask HN: Any Local LLM can I run without GPU for Local Agentic workflow AI?
    8 points by limondas 28 days ago | 4 comments
23. Show HN: Selora – local model for Home Assistant (github.com)
    7 points by bayshark 21 days ago | 4 comments
24. I patched llama.cpp to gain 20% prompt processing TPS. Help me make a PR
    6 points by i_am_rocoe 11 days ago | 2 comments
25. Ask HN: How is GPU power draw measured at scale?
    6 points by anax32 12 days ago | 2 comments
26. As; HN: I was curious why MTP affects PP TPS in llama.cpp. My PoC recovers it?
    4 points by i_am_rocoe 13 days ago | 1 comments
27. Ask HN: What are some good/fast coding models for Apple Silicon?
    2 points by LoganDark 22 days ago | 3 comments
28. Show HN: Apex-1-flash, 4B LLM finetuned on RTX 5070 (huggingface.co)
    2 points by Qmay_Dev 11 days ago | discuss
29. Ask HN: What do you do when you hit Claude subscription limits?
    2 points by sermakarevich 26 days ago | discuss
30. Ask HN: Is it just me or does Claude / Sonnet 5 sound condescending recently?
    1 points by alentred 5 days ago | 2 comments