Top all-time · #4
OpenAI News
Company blog · openai.comai
OpenAI News is the company blog at openai.com, writing about ai. Since 2015, 2,220 different Hacker News readers have linked it 3,845 times, most often “Starting Sep 1, GPT-3 is becoming 3X Cheaper”. It is still writing: the last post was on 2026-11-02, and you can follow it by feed.
Rarely recommended · 1 of 36 sampled comments recommend it?
Why HN links it
HN readers cite OpenAI's news pages for model launches, Sora, GPT-4o, safety and policy statements, and pricing or benchmark claims, mostly as sources for their own points rather than as judged writing.
- Recommended for · 1 comments
- The GPT-6 Astra announcement is called much better presented than a Gemini 4 Argon announcement, with every metric shown as better.1
Summarised by Claude from a sample of 36 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- Starting Sep 1, GPT-3 is becoming 3X Cheaper42 people · 47 mentions · first linked 2021-12-14
- Microsoft is a minority owner of OpenAI Global LLC41 people · 52 mentions · first linked 2023-11-17
- /charter/36 people · 45 mentions · first linked 2019-03-11
- Harness Engineering32 people · 32 mentions · first linked 2026-02-11
- GPT-430 people · 36 mentions · first linked 2023-03-14
- Clip: Connecting Text and Images28 people · 32 mentions · first linked 2021-01-05
- Better Language Models and Their Implications25 people · 28 mentions · first linked 2019-03-25
- GPT4 is up to 6 times more expensive than GPT3.525 people · 27 mentions · first linked 2023-03-08
- OpenAI ChatGPT: Optimizing Language Models for Dialogue24 people · 28 mentions · first linked 2022-11-30
- New OpenAI Privacy Policy and TOS24 people · 24 mentions · first linked 2023-02-28
What Hacker News says?
On the other hand, this is the first convincing use of a “diffusion transformer” [1]. My understanding is that videos and images are tokenized into patches, through a process that compresses the video/images into abstracted concepts in latent space. Those patches (image/video concepts in latent space) can then be used with transformers (because patches are the tokens). The point is that there is plenty of room for o…
Comment linking /research/video-generation-models-as-world-simulators
on “Sora: Creating video from text”Submitted Sora: Creating video from text as a story · 3647 points
HN discussion> It just computes next frame based on current one and what it learned before, it's a plausible continuation. ... > Now it's just a "what is the next frame of this 2D image" model plus some textual context. This is incorrect. Sora is not an autoregressive model like GPT, but a diffusion transformer. From the technical report[1], it is clear that it predicts the entire sequence of spatiotemporal patches at once. [1]:…
Comment linking /research/video-generation-models-as-world-simulators
on “Sora: Creating video from text”Submitted GPT-4o as a story · 3138 points
HN discussionTechnical report here: https://openai.com/research/video-generation-models-as-world...
Comment linking /research/video-generation-models-as-world-simulators
on “Sora: Creating video from text”Submitted GPT-6 Astra as a story · 2279 points
HN discussionAlphaGo and AlphaZero were able to achieve superhuman performance due to the availability of perfect simulators for the game of Go. There is no such simulator for the real world we live in. (Although pure LLMs sorta learn a rough, abstract representation of the world as perceived by humans.) Sora is an attempt to build such a simulator using deep learning. This actually affirms my comment above. “Our results suggest…
Comment linking /research/video-generation-models-as-world-simulators
on “Sora: Creating video from text”Submitted Open models by OpenAI as a story · 2124 points
HN discussionThe Lagos video ( https://openai.com/sora?video=lagos ) is very much how my dreams unfold. One moment, I'm with my friends in a bustling marketplace, then suddenly we are no longer at the marketplace, but rather overlooking a sunset and a highway. I wonder if there are some conceptual similarities how dreams and AI video models work.
Comment linking /sora?video=lagos
on “Sora: Creating video from text”- rd · 2025-08-07
Submitted GPT-5 as a story · 2063 points
HN discussion https://openai.com/sora?video=cat-on-bed Even though many things are super impressive, there is a lot of uncanny valley happening here.
Comment linking /sora?video=cat-on-bed
on “Sora: Creating video from text”Submitted Dall-E 2 as a story · 1820 points
HN discussionThis is bananas. This is ahead of anything else I've seen. The entire stock footage industry may be shut down over night because of something like this. And it is still not perfect. Looking at the example of the plastic chair being dug up in the desert[1] is frankly a bit... funky. But imagine in 5 or even 10 years. 1. https://openai.com/sora?video=chair-archaeology
Comment linking /sora?video=chair-archaeology
on “Sora: Creating video from text”Submitted GPT-6 Sol and Luna as a story · 1779 points
HN discussionTotally agree, they can pay a lot of monkeys at typewriters, but also? https://openai.com/sora?video=big-sur made me literally say, out loud, "doesnt-matter-had-sex"
Comment linking /sora?video=big-sur
on “Sora: Creating video from text”Submitted Whisper – open source speech recognition by OpenAI as a story · 1705 points
HN discussionhttps://openai.com/sora?video=big-sur In this video, there's extremely consistent geometry as the camera moves, but the texture of the trees/shrubs on the top of the cliff on the left seems to remain very flat, reminiscent of low-poly geometry in games. I wonder if this is an artifact of the way videos are generated. Is the model separating scene geometry from camera? Maybe some sort of video-NeRF or Gaussian Splatt…
Comment linking /sora?video=big-sur
on “Sora: Creating video from text”Submitted Learning to Reason with LLMs as a story · 1654 points
HN discussionAm I using it wrong? I have the gpt plus subscription, and can select "gpt4o" from the model list on ChatGPT, but whichever example I try from the example list under "Explorations of capabilities" on ` https://openai.com/index/hello-gpt-4o/ `, my results are worse: * "Poetic typography" sample: I paste the prompt, and get an image with the typical lack of coherent text, just mangled letters. * "Visual Narratives: Ro…
Comment linking /index/hello-gpt-4o/
on “GPT-4o”Submitted ChatGPT Search as a story · 1485 points
HN discussionhttps://openai.com/index/memory-and-new-controls-for-chatgpt...
Comment linking /index/memory-and-new-controls-for-chatgpt/
on “GPT-4o”Submitted Sharing AI progress in mathematics as a story · 1363 points
HN discussionjust need the helmet https://openai.com/index/be-my-eyes/
Comment linking /index/be-my-eyes/
on “GPT-4o”Submitted On the Navier–Stokes Millennium Prize Problem as a story · 1346 points
HN discussionJust found there's a whole section about that in this post: https://openai.com/index/hello-gpt-4o/ It says "Japanese 1.4x fewer tokens (from 37 to 26)" - some other languages get much bigger improvements though, best is "Gujarati 4.4x fewer tokens (from 145 to 33)".
Comment linking /index/hello-gpt-4o/
on “GPT-4o”Submitted DALL·E: Creating Images from Text as a story · 1266 points
HN discussionIt's twice as fast/cheap relative to GPT-4-turbo, which is still expensive compared to GPT-3.5-turbo and Claude Haiku. https://openai.com/api/pricing/
Comment linking /api/pricing/
on “GPT-4o”Submitted GPT-5.2 as a story · 1195 points
HN discussionI can't see any calculator for the audio pricing ( https://openai.com/api/pricing/ ) or document type field in the Chat Completions API ( https://platform.openai.com/docs/api-reference/chat/create ) for this new model. Is the audio in API not available yet?
Comment linking /api/pricing/
on “GPT-4o”Submitted Three new discoveries in mathematics, computer science, and physics as a story · 1161 points
HN discussion
