دليل نموذج لغة الذكاء الاصطناعي من الجيل التالي ChatGPT o1

This is amazing. Used chatgpt o1-preview to create an animation of a spinning donut in my terminal. I'm switching back to chatGPT from claude. This is the first try btw #ChatGPT #strawberry pic.twitter.com/P33nywISq6

— Jordan (@schtwiller) September 12, 2024

ChatGPTから新しいモデルのo1がリリース。まとめてみました👇
- 博士課程の学生と同等のパフォーマンス(物理、化学、生物など)
- 数学とコーディングのタスクにも優れる
- 同時にo1-miniもリリース(コーディング特化)
- 特化型であり全てのタスクでGPT-4oを上回るわけではない… pic.twitter.com/kFsTiI5rTW

— 矢野 哲平 | 耳で学ぶChatGPT (@robothink_jp) September 12, 2024

Okaaay. Die Preview von ChatGPT o1 ist wirklich beeindruckend. Ein kurzer Prompt und man erhält z.B. ein funktionsfreies Spiel. Erweiterungen/Änderungen werden korrekt vorgenommen. Hab mal Space Invaders nachbauen lassen als Browsergame: pic.twitter.com/BN4sGJ56bb

— Hendrik (@ginandstocks) September 13, 2024

Gave Claude 3.5 Sonnet and ChatGPT o1 this physics problem.

Claude gave the wrong answer 3 consecutive times.

o1 nailed it in the first attempt and the chain of thought was pretty impressive - exactly how I attempted to solve it.

Impressive so far! pic.twitter.com/J4O3bgZGVv

— Abhishek Agarwal (e/acc) (@abhi_agarwal4) September 12, 2024

This is wild.

OpenAI just launched "o1" models in ChatGPT.

These models are capable of PhD-level reasoning.

Now ChatGPT "Thinks" before responding.

12 wild examples of o1 models: 👇 pic.twitter.com/DnHKTSSdFT

— Ihtesham Haider (@ihteshamit) September 13, 2024

Test #3 of ChatGPT o1:

Write a story starting each word with the letter A to Z, without ever repeating the same letter.

Holy moly... it did it. pic.twitter.com/gA3q8iCR0n

— Ruben Hassid (@RubenHssd) September 12, 2024

ChatGPT Just Got Smarter

ChatGPT o1-preview showcases deep reasoning in real time.

People often believe fast responses mean smarter AI.

However, AI that takes time to break down your queries leads to far more insightful answers.

Here are 3 surprising benefits we’ve… pic.twitter.com/NNXZW2ZkMl

— Clintin Lyle Kruger (@Lyle_AI) September 12, 2024

ChatGPT o1 is much better than 4o for very simple basic reasoning at least. pic.twitter.com/WYW4Hd415Z

— Damien Henry (@dh7net) September 13, 2024

Someone tell Andrej that o1-mini can cook with the Riemann Hypothesis. Prompt here:https://t.co/sqtAPaHz0Z pic.twitter.com/cP1uGqMqlC

— Ray Fernando (@RayFernando1337) September 13, 2024

ChatGPT 4o VS o1-preview

Simple video showing how good o1-preview is at understanding and logic.#openai #o1preview pic.twitter.com/AxBjwUkxjo

— Fonts Keyboard AI (@FontsKeyboardAI) September 12, 2024

Did a quick test to see the capability of the new @OpenAI ChatGPT o1-preview on its ability to code and I have to say that it's doing a pretty good job. Had to tell it to adjust a few things to work with the size of the images and white space around it, etc, and also animation… pic.twitter.com/65QKTeZAtm

— Jack Vinijtrongjit | Saakuru Labs (@juminoz) September 13, 2024

Pretty sick how ChatGPT, with o1, become so much better with added reasoning.

You can see what it's doing. I think a lot of the AI's will become much smarter with this type of improvement without improving the "actual" AI.https://t.co/ido6p6FRFe pic.twitter.com/rv3KshGHQv

— adriaan.com 📊 Simple Analytics (@adriaandotcom) September 13, 2024

I asked ChatGPT o1 to reconstruct Mill's argument for the principle of utility so that it contains no logical flaws but only hard-to-defend premises. Here is what it generated almost immediately. pic.twitter.com/Xu2AfPSIMi

— Kevin Vallier (@kvallier) September 12, 2024