Hackernews posts about arXiv.org

ArXiv.ORG is an online repository that provides open access to electronic preprints of completed peer-reviewed papers in physics, mathematics, computer science, and related disciplines.

1. Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots (cactuscompute.com)
   517 points by HenryNdubuaku 1 day ago | 173 comments
2. Show HN: I simulated closing the Strait of Hormuz on real oil trade data (globaloilnetwork.staffinganalytics.io)
   244 points by eliotho 20 days ago | 128 comments
3. Show HN: Optimize and serve models with Fable quality at half the cost (github.com)
   71 points by SilenN 16 days ago | 30 comments
4. Show HN: Inflexa – open-source Intelligence for Biology (github.com)
   7 points by tanasaradu 22 days ago | 2 comments
5. Grok is a surprisingly good automated theorem prover
   3 points by henryrobbins00 21 days ago | 1 comments
6. Show HN: Running over 80M tokens in one agent session with no compaction (github.com)
   3 points by andrew_ocs 29 days ago | discuss
7. Weird password rules for website to register
   2 points by guptadeepak 28 days ago | 5 comments
8. Making models smarter and cheaper at the same time
   2 points by Wetime 23 days ago | 1 comments
9. Show HN: Run SAM only when your tracker is uncertain – #1 on SportsMOT (www.holma.io)
   2 points by lapuerta 29 days ago | discuss
10. Show HN: DFAH-Bench – same agent decision, different tool paths (github.com)
    1 points by raffisk 10 days ago | discuss
11. Handbook.md shows that long policy documents do not reliably govern agents (arxiv.org)
    325 points by spIrr 14 days ago | 209 comments
12. Kimi Linear: An Expressive, Efficient Attention Architecture (2025) (arxiv.org)
    306 points by ronfriedhaber 15 days ago | 133 comments
13. Mathematics of Data Science (arxiv.org)
    222 points by Anon84 27 days ago | 14 comments
14. Sycophantic AI Decreases Prosocial Intentions and Promotes Dependence (2025) (arxiv.org)
    174 points by robin_reala 7 days ago | 107 comments
15. The Maxwell Conjecture Is False (GPT 5.6 Sol) (arxiv.org)
    157 points by rahen 12 days ago | 142 comments
16. Why Large Language Models Fail at Tabular Prediction (arxiv.org)
    115 points by sbulaev 8 days ago | 33 comments
17. The Tragedy of the Cognitive Commons (arxiv.org)
    110 points by jmintz 2 days ago | 77 comments
18. Zigbee vs. Matter over Thread:Understanding IoT Protocol Performance in Practice (arxiv.org)
    106 points by teleforce 7 days ago | 102 comments
19. Harmony Explained: Progress Towards a Scientific Theory of Music (2012) (arxiv.org)
    105 points by surprisetalk 15 days ago | 114 comments
20. When AI Benchmarks Plateau: A Systematic Study of Benchmark Saturation (arxiv.org)
    104 points by doppp 8 days ago | 131 comments
21. Zero-Mem: Zero-Token Memory Operations for LLM Agents (arxiv.org)
    101 points by theanonymousone 7 days ago | 13 comments
22. AI migrated legacy COBOL programs to Java, bugs included (arxiv.org)
    99 points by felineflock 9 days ago | 99 comments
23. How real are real numbers? (2004) (arxiv.org)
    96 points by surprisetalk 16 days ago | 103 comments
24. Coding agents think ahead of time (arxiv.org)
    96 points by andre15silva 29 days ago | 78 comments
25. Co-evolution of self-replication and function in a digital primordial soup (arxiv.org)
    92 points by vicgalle_ 25 days ago | 35 comments
26. Qubes OS Security in the Public Record (arxiv.org)
    92 points by sciences44 25 days ago | 18 comments
27. Can LLMs Perform Deep Technical Comprehension of Computer Architecture Papers (arxiv.org)
    83 points by Jimmc414 27 days ago | 26 comments
28. A taxonomy of omnicidal futures involving artificial intelligence (2025) (arxiv.org)
    78 points by amelius 20 days ago | 83 comments
29. k-Coloring is Faster than Computing the Chromatic Number (arxiv.org)
    73 points by matt_d 12 days ago | 23 comments
30. A Study of Microsoft's Early 2026 Rollout of Claude Code and GitHub Copilot CLI (arxiv.org)
    71 points by softwaredoug 30 days ago | 41 comments