EN

GPT-5 Review: The Truth After a Week of Use

#AI Tools
2025-09-02 06:2710 min read

Content Introduction

In this video, the presenter reviews OpenAI's new flagship model, GP5, sharing mixed feelings after a week of daily use. They discuss its capabilities, highlighting the multimodal functionality, which allows the model to process both text and images. While GP5 excels at advanced tasks like creativity and coding, it struggles with basic tasks. The video emphasizes its context window limitations and parameter count uncertainty, comparing it unfavorably with other models for specific use cases. The presenter concludes by suggesting that GP5 is best for complex problem-solving and coding, while recommending alternatives for simpler tasks. Ultimately, they emphasize the need for OpenAI to balance simplicity and specialization in future updates.

Key Information

  • GP5 is the new flagship model from OpenAI, which has elicited mixed feelings from users.
  • The model has been tested for a week, handling a variety of tasks including coding, research, creativity projects, and daily productivity.
  • While it excels at complex tasks, GP5 struggles with basic operations, leading to user frustrations.
  • GP5 is multimodal, accepting text and images as input, and operates as a system of specialized models rather than a single massive model.
  • The model’s key flaw is in its decision-making regarding which tool to use, affecting user experience.
  • GP5 has a context window of about 200,000 tokens, which is significantly less than the state-of-the-art capacity of 1 million tokens.
  • User experience with GP5 varies; it performs exceptionally well with coding tasks and creative projects but has issues with tasks involving sensitive communications.
  • The model's simplifications may limit user control over specific tasks, creating dissatisfaction.
  • Despite its downsides, users find value in GP5 for coding and creativity, indicating potential if OpenAI can balance simplicity with specialization in future updates.

Timeline Analysis

Content Keywords

GPT5

GPT5 is the new flagship model from OpenAI, which the speaker has mixed feelings about. It excels at complex problem-solving, coding, and creativity, but struggles with basic tasks and email drafting.

Multimodal Input

GPT5 accepts both text and images as input, making it a versatile tool. However, it is built from a system of specialized models rather than one massive model.

Performance Issues

Despite its strengths in advanced tasks, GPT5 is frustratingly poor at basic tasks. The single feature of determining the best model for a job is also seen as a flaw.

Context Window

The context window of GPT5 allows it to remember roughly 200,000 tokens, but this is less than the state-of-the-art standard of 1 million tokens.

Daily Use Cases

The speaker frequently uses GPT5 for writing tasks like drafting emails, but finds that it often alters meanings instead of keeping them intact.

Coding and Creativity

GPT5 excels particularly in coding, allowing for remarkable creations like fluid dynamics simulations and UI refactoring. The model provides significant help in creative projects.

User Experience

Using GPT5 feels simplified to a single choice of model for tasks, potentially reducing user freedom compared to previous versions where users could choose from various model options.

Safety and Security

The speaker notes that they have not been able to jailbreak GPT5 for unsafe instructions or content, which aligns with their safety concerns regarding AI.

Comparison with Other Models

For tasks requiring sensitive communication, the speaker suggests using other models like Gemini 2.5 or GPT-4 rather than GPT5.

Future of GPT

The model's future usefulness will depend on OpenAI's ability to balance simplicity with specialization in its AI offerings.

Related questions&answers

What is GP5?

GP5 is the new flagship model from OpenAI that has received mixed reviews.

What has the reviewer been using GP5 for?

The reviewer has been using GP5 as their daily driver for tasks like coding, research, creativity projects, and daily productivity.

What are the strengths of GP5?

GP5 excels at complex tasks and coding but struggles with basic tasks.

How is GP5's architecture different from previous models?

GP5 is not just a single massive model, but a system of specialized models that can accept both text and images as input.

What is a significant flaw of GP5?

A significant flaw is its context window, which is limited to roughly 200,000 tokens, whereas the state-of-the-art is around 1 million tokens.

What has been the reviewer's experience with writing using GP5?

The reviewer found that GP5 often alters the meaning of their writing, making it frustrating to use for drafting emails and other written communications.

What changes were noted in multi-turn conversations?

Multi-turn conversations have improved, with GP5 showing the ability to get back on track after confusion.

Is GP5 effective for safety and security?

The reviewer hasn't found success in jailbreaking GP5, noting that it does not produce unsafe instructions or political propaganda.

Who is GP5 recommended for?

GP5 is recommended for users needing AI for complex problem-solving, planning, creativity, and especially coding.

What are the overall recommendations for using GP5?

If sensitive communication is needed, it's suggested to use alternatives like Gemini 2.5 Flash or GPT-4 instead.

What does the reviewer think about the future of GP5?

The reviewer believes that if OpenAI can provide both simplicity and specialization, GP5 could be successful in the future.

More video recommendations

Share to: