Hackernews posts about GPT4

GPT4 is a highly advanced language model designed by Meta AI that has surpassed the capabilities of its predecessors in terms of processing power and ability to generate human-like text.

1. Ask HN: How can ChatGPT serve 700M users when I can't run one GPT-4 locally?
   574 points by superasn 26 days ago | 365 comments
2. From GPT-4 to GPT-5: Measuring progress through MedHELM [pdf] (www.fertrevino.com)
   127 points by fertrevino 12 days ago | 96 comments
3. Show HN: I fine-tuned GPT4.1 on my iMessage history (jonyork.net)
   10 points by jonpizza 6 days ago | 6 comments
4. ChatGPT 5 power consumption could be as much as eight times higher than GPT 4 (www.tomshardware.com)
   6 points by indigoabstract 18 days ago | 1 comments
5. Configuring GH Codespaces with UV/node + llm tool + free GPT4.1 w/$GITHUB_TOKEN (til.simonwillison.net)
   4 points by indigodaddy 21 days ago | discuss
6. Bill Gates does not expect GPT-5 to be much better than GPT-4 (2023) (the-decoder.com)
   3 points by geox 19 days ago | discuss
7. GPT-5 vs. GPT-4 for AI medical diagnosis examples (github.com)
   3 points by jph 26 days ago | discuss
8. When asked what model it is, GTP-5 says it's GPT-4.5
   2 points by jari_mustonen 26 days ago | 2 comments
9. Small LLMs can outperform GPT-4 at detecting jailbreaks (romaingrx.com)
   2 points by romaingrx 26 days ago | 1 comments
10. I Co-Wrote a Modern Prophetic Trilogy with GPT-4: Here's What Emerged (www.scribd.com)
    2 points by nickprophet 29 days ago | 1 comments
11. GPT-OSS 20B delivers GPT-4.1 scores at 5x the speed (www.dbreunig.com)
    2 points by dbreunig 28 days ago | discuss
12. Ask HN: How did you like GPT-4.5?
    1 points by felipemesquita 27 days ago | 2 comments
13. Free AI large model API, gpt4o0.02 per piece (grsai.com)
    1 points by AllenRen 16 days ago | 1 comments
14. Notes on Image Generation with GPT-4.1 (taoofmac.com)
    1 points by rcarmo 14 days ago | discuss
15. Launch HN: Dedalus Labs (YC S25) – Vercel for Agents
    77 points by windsor 6 days ago | 15 comments
16. Show HN: Happy Coder – End-to-End Encrypted Mobile Client for Claude Code (github.com)
    30 points by ex3ndr 20 days ago | 8 comments
17. Show HN: Dinoki – Privacy-First Desktop AI with Pixel Pets (6MB Native)
    17 points by tpae 29 days ago | 7 comments
18. GPT5 is worse than 4.1-mini for text and worse than Sonnet 4 for coding
    10 points by hitradostava 24 days ago | 16 comments
19. Show HN: Pluely – open-source alternative to Cluely (~10MB, Always On Display) (github.com)
    7 points by truly_sn 15 days ago | 5 comments
20. Show HN: GPT-5 Document Retrieval – AI Assistant with Inline Citations (www.smartresearch-ai.com)
    5 points by ben011 26 days ago | discuss
21. Ask HN: OpenAI GPT-5 API seems to be significantly slower – is this expected?
    4 points by tlogan 25 days ago | 3 comments
22. Show HN: I spent 40 hours vibe coding after 40 years of traditional programming (marcobenedetti.substack.com)
    3 points by maxbene 27 days ago | discuss
23. Modular LLM framework inspired by Linux – aiming for a one-GPU future
    2 points by openkame 8 days ago | 1 comments
24. Show HN: I built an open-source CSV importer that I wish existed (github.com)
    2 points by aray07 4 days ago | discuss
25. Show HN: Llmswap v3.0 – CLI and SDK for OpenAI, Claude, Gemini, Watsonx (pypi.org)
    2 points by sreenathmenon 14 days ago | discuss
26. Ask HN: What is "response-level error rate" and how is it measured?
    2 points by myyke 26 days ago | discuss
27. Show HN: LLM Opinion – discover what sources an LLM relies on (llm-opinion.onrender.com)
    2 points by piaxar 28 days ago | discuss
28. GPT5 is the best coding LLM because other LLMs admit it?
    1 points by adinhitlore 8 days ago | 8 comments
29. Ask HN: What are you expecting from the GPT-5 later?
    1 points by darajava 27 days ago | 5 comments
30. Ask HN: Would you use a chat-based money tracker?
    1 points by chetansorted 21 days ago | 3 comments