OpenAI Developer Day: New Features for GPT-4 and Mini

### New Features from OpenAI Developer Day
OpenAI is launching new features for developers, with the GPT-4 and Mini versions offering **similar performance at a 20 times lower price**. The Mini version provides a **larger context window of about 16,383 tokens**, making it more favorable for writing longer texts. The Playground now allows users to **write system instructions**, facilitating the creation of AI agents.

### Creative Writing Expert Agent and GPT-4 Benefits
The generator can be set as a **creative writing expert agent**, optimizing tasks for users. It offers features like webtoon content generation, including **titles, episode summaries, and target audience descriptions**, at a low cost. OpenAI has expanded API usage limits for developers, increasing **request handling capacity** for GPT-4 models.

### Updates on GPT-4 and Its Impact
GPT-4's availability has been extended, with its performance now **matching or exceeding that of GPT-5**. The model's response time has increased, suggesting a **more complex thinking process**. OpenAI is testing new models exclusively through GPT-4, leading to **higher resource consumption and costs**.

### The Impact of GPT-4's New Features
GPT-4 has introduced additional functions, such as **image editing**, and is now providing responses in a testing format labeled as **1 and 2**. The model's thinking function, which offers this testing approach, is being incorporated into its responses for **all questions**. The frequency of these previews is high, with paid subscribers receiving around **50 times a day**, potentially leading to the inference function being offered **more frequently in the future**.

### Improving Pony's Reasoning Abilities
Pony's reasoning abilities have **significantly improved** in the past month and a half. The Orion project, part of the Q and Strawberry projects, will integrate the reasoning model into GPT-4, **activating it for complex questions**. The output model's gradual distribution aims to **enhance reasoning capabilities** and **advanced data analysis features**.

OpenAI's new features enhance GPT-4 Mini, offering similar performance to GPT-4 at a fraction of the cost.

### OpenAI Developer Day: New Features and Insights on GPT-4 and Mini
* 🌅 Today, at the OpenAI Developer Day in San Francisco, several new features were launched that will **enhance the developer experience** on the platform.
* 🛠 The performance of GPT-4 and Mini is nearly identical, yet Mini is available at a **20 times lower cost**, making it an attractive option for users.
* 📜 GPT-4 on Mini allows for a context window of approximately **16,383 tokens**, which is beneficial for creating longer texts, while the standard GPT-4 has a maximum of **4,095 tokens**.
* 💡 Users are encouraged to utilize the free version of GPT-4 on Mini via the API Playground, which offers capabilities for **creative writing without incurring significant costs**—around **$0.01 per token**.
* 🚀 OpenAI has also expanded API usage limits, with GPT-4 Mini now capable of handling **30,000 requests per minute**, which greatly benefits developers.
* 🔍 The announcement highlighted that GPT-4 is undergoing **significant changes**, with **improved reasoning capabilities** expected in future updates, particularly with the integration of the **Orion project**.
* 📈 As the output model evolves, users can anticipate **better responses and enhanced reasoning abilities**, particularly for **complex questions**, which is promising news for the AI community.
* 📽 This video was created to **urgently share these developments**, and more exciting updates are expected to follow.

LiveWiki

GPT4o가 생각하기 시작 | 누구나 AI Agent를 만들 수 있다 | OpenAI 개발자 데이 |

GPT4o가 생각하기 시작 | 누구나 AI Agent를 만들 수 있다 | OpenAI 개발자 데이 |