New ChatGPT 4.5 is Here - The Good, The Bad and The UGLY

2025-03-13 12:118 min read

Content Introduction

In this video, the presenter discusses their experience with OpenAI's newly released GPT-4.5, following nearly a day of testing. They compare GPT-4.5 against other models, including the CLAE 3.7 and older ChatGPT versions. They highlight the current access limitations and advise against upgrading to the $200/month plan until it's more established. The video emphasizes the improvements in Q&A accuracy and reduced hallucination rates but also notes the shortcomings in reasoning capabilities. The presenter showcases various tests, demonstrating GPT-4.5's strengths in emotional intelligence and writing, while also revealing its slower response times compared to competitors. Overall, the video provides insights into the practical applications and limitations of GPT-4.5, encouraging viewer feedback and interaction.

Key Information

  • OpenAI has released GPT 4.5, which the speaker has been testing and comparing against other models like CLAE 3.7 and the previous chat GPT model.
  • Access to GPT 4.5 is currently limited to a $200 a month plan, but upcoming changes will include availability in different plans such as Education and Enterprise.
  • The speaker found that GPT 4.5 has improvements in Q&A accuracy, broader knowledge, and lower hallucination rates compared to previous models, but it is still in research preview.
  • Although GPT 4.5 is marketed as the best model for chat, it is not a reasoning model, which limits its performance on certain benchmarks.
  • Tests showed that GPT 4.5 is better at emotional intelligence tasks and generating concise answers, but users might find slower response speeds compared to older models.
  • The speaker expressed confusion over the pricing model of the new release and noted that many features of GPT 4.5 may not significantly differentiate it from previous versions, particularly in speed and reasoning capabilities.
  • Users generally might not see a huge upgrade in functionalities, which leads to skepticism about whether the model justifies the price increase.

Timeline Analysis

Content Keywords

GPT 4.5

OpenAI has released GPT 4.5, which has been tested against other models including CLAE 3.7 and the older chat GPT model. The early access to this version is currently available through a $200 a month plan.

User Access

GPT 4.5 is still in research preview and will soon be accessible through different plans, including education and enterprise plans, designed to accommodate various needs.

Model Confusion

The introduction of many AI models has led to confusion among users, highlighting the need for a simpler model selection process.

Model Picker

Future updates may include a model picker feature to simplify the use of different models for various tasks without needing to choose manually.

Q&A Accuracy

GPT 4.5 reportedly shows improvements in simple Q&A accuracy, achieving higher scores than previous models, while also demonstrating reduced hallucination rates.

Hallucination Rates

GPT 4.5 exhibits lower hallucination rates compared to its predecessors, making it a more reliable choice for generating accurate information.

Emotional Intelligence

GPT 4.5 has been noted for its improved emotional intelligence, making it more capable of understanding and responding to human-like contexts interactively.

Practical Use Cases

Examples of practical applications for GPT 4.5 include customer service automation, creating sincere messages for employees, and providing detailed assistance in business operations.

Testing Applications

A comparison was made between GPT 4.5 and other models like CLAE 3.5 in terms of speed and reliability when handling prompts and tasks.

Pricing Structure

Concerns were raised regarding the pricing structure of OpenAI's GPT models, with comparisons made to previous models highlighting accessibility issues for developers.

More video recommendations