Hackernews posts about Model 3/Y

1. Gemini 3.8 Flash and 3.8 Flash Cyber (blog.google)
   76 points by simonsarris 18 days ago | discuss
2. Show HN: PicoLM v1.0-rc1 (github.com)
   3 points by gabucino 17 days ago | discuss
3. Show HN: CleanMyMedia – organize decades of photos, with local face recognition (cleanmymedia.com)
   3 points by ed-cmm 18 days ago | discuss
4. Show HN: LLM-powered webapp to build LLM-powered webapps (codeplusequalsai.com)
   2 points by cryptoz 26 days ago | 2 comments
5. Show HN: What differentiates developers when we all use AI? (aicalculatorpro.com)
   2 points by helloaiworld 19 days ago | discuss
6. Show HN: I might have built the best Linear alternative (github.com)
   1 points by Croffasia 22 days ago | discuss
7. Ask HN: What is the best local model that runs on your Mac at decent speed?
   1 points by carloslfu 4 days ago | discuss
8. Show HN: BAIhAIs – an autonomous art school for AI agents (baihais.com)
   5 points by demonstrandom 24 days ago | 2 comments
9. Show HN: Use GLM-5.3 in Cursor today via tokengo API (www.tokengo.com)
   3 points by bingus-bongo 25 days ago | 2 comments
10. Show HN: ForgeGuardian – Open-source software supply-chain security scanner (github.com)
    3 points by mahi1337 15 days ago | 1 comments
11. Tell HN: An inside view of Montana's new biotech law
    34 points by niklas_anzinger 4 days ago | 25 comments
12. I've tested some local LLMs on prosumer hardware, here are some findings
    7 points by felineflock 28 days ago | discuss
13. Show HN: Ramanujan-computing: use idle computation to run scientific computing (github.com)
    3 points by ps_ramanujan 23 days ago | discuss
14. Ask HN: What's the most economical approach to the most tokens?
    2 points by RufusJacksons 4 days ago | discuss
15. Show HN: Replay – Audit silent prompt cache misses in AI agent transcripts (replay.doctor)
    1 points by danielsaito 7 days ago | discuss
16. Show HN: Cactus Needle 3: 8-29MB automation models can match DeepSeek V4 Flash (cactuscompute.com)
    230 points by HenryNdubuaku 3 days ago | 91 comments
17. Harness your expectations: a 27B model matched GLM-5.3-Flash after leak fixes (aistack.imec-int.com)
    3 points by flifenstein 5 days ago | discuss
18. Show HN: EACL is a situated ReBAC authorization library for Datomic and Datahike (github.com)
    5 points by pgt 5 days ago | discuss
19. Ask HN: Anyone care to answer a small survey about ML?
    1 points by aunekiaru 21 days ago | 2 comments
20. Show HN: ThunderPhone v2 – a new architecture for voice AI (thunderphone.com)
    4 points by kolchinski 24 days ago | discuss
21. Show HN: I made an 8 bit game that teaches beginners to pwn basic C programs (github.com)
    2 points by udbhavs 5 days ago | discuss
22. Show HN: Open-source simulation testing infra for voice agents (github.com)
    16 points by Nischalj10 10 days ago | 7 comments
23. Ask HN: AI writes better code than me. How to keep my identity?
    15 points by jdw64 23 days ago | 25 comments
24. Is Google planning for Gemini 4 rather than 3.5 pro?
    5 points by Qhloi 11 days ago | 1 comments
25. Show HN: MultiMatte, a Promptable Image Background Removal Model (usefeyn.com)
    57 points by snyy 10 days ago | 8 comments
26. Launch HN: Hebbian Robotics (YC S26) – Build scalable robotics data pipelines (github.com)
    46 points by kstonekuan 20 days ago | 12 comments
27. Show HN: Swift-Qwen3.8-27B, -58.3% thinking, x1.95 speed, accuracy of xhigh (huggingface.co)
    30 points by kisjovan 4 days ago | 17 comments
28. Show HN: Analyst Index – analysts who make money telling you good stock calls (www.analystidx.com)
    15 points by haichuan 7 days ago | 15 comments
29. Show HN: Benzi – Code Intelligence Infrastructure for Frontier AI Models (github.com)
    3 points by tweedler290 13 days ago | discuss
30. Show HN: Train 300M/32-Layer Model in 1.5GB RAM on Base M1 Mac
    3 points by vlad_kalinkin 21 days ago | discuss