GPT-4.5 sur ChatGPT : Mon Test Complet (Top ou Flop ?)

GPT-4.5 sur ChatGPT : Mon Test Complet (Top ou Flop ?)

by Ludo Salenne

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQFGenerative artificial intelligence
🎙 Ludo Salenne 👥 267K 📅 February 28, 2025 ⏱ 38 min 👁 56K 📄 expert opinion 🧭 2026-08-21
Available in: English (current) Français

Keywords

GPT-4.5ChatGPTtestcomparisonAI

Summary

The video is a comprehensive test of GPT-4.5, recently released on ChatGPT. The creator, Ludo Salenne, begins by explaining what GPT-4.5 is, highlighting its improved natural interactions, better intention understanding, reduced hallucinations, and enhanced multilingual performance. He then conducts several tests: a first emotional prompt to assess empathy, a comparison with GPT-4o using a ‘genius of evil’ prompt, a coding test against Claude 3.7, and tests of search, PDF analysis, and vision capabilities. Throughout, he notes GPT-4.5’s more concise and precise responses, its better role-playing adherence, and its impressive performance in various tasks. He also discusses access (initially Pro users, then Plus/Team) and mentions the official system card. The video concludes with his overall opinion, praising the model’s improvements but noting it is not a revolutionary leap. The presentation is energetic and practical, aimed at a general audience interested in AI applications.

142 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides valuable hands-on demonstrations of GPT-4.5’s capabilities, offering concrete examples of its performance in emotional understanding, creative prompting, coding, and multimodal tasks. The argumentation is based on personal testing and comparisons with other models, which gives practical insights but lacks systematic methodology. The creator’s enthusiasm is evident, but conclusions are subjective and may not be generalizable. The comparisons with GPT-4o and Claude 3.7 are illustrative but not exhaustive, and the evaluation criteria are not clearly defined. Overall, the information is useful for users considering GPT-4.5, but the argumentation is more anecdotal than rigorous.

Scientific Rigor, Source Quality, Title Accuracy

The video references the official OpenAI system card for GPT-4.5, providing a credible source for technical details. However, most claims are based on the creator’s own tests, which are not peer-reviewed or systematically designed. The title accurately reflects the content, and the video includes a clear structure with chapters. The creator also mentions other videos and resources, but these are promotional rather than scientific. The lack of external validation and the subjective nature of the tests reduce the overall scientific rigor. The video is more of an expert opinion than a rigorous evaluation.

203 words

Title / Content Match

The title accurately reflects the content: a comprehensive test of GPT-4.5 on ChatGPT, covering strengths, comparisons, and verdict.

Quality & Reliability

6/10

The video is a hands-on test of GPT-4.5, with subjective comparisons and limited methodological rigor. The creator references the official system card and provides practical demonstrations, but conclusions are based on personal experience rather than systematic evaluation.

Chapters

Cited Sources

  • GPT-4.5 System Card — Official technical report on GPT-4.5, referenced for details on training and capabilities.
  • Trustpilot Reviews — Customer reviews for the creator's training, mentioned as social proof.
  • Formation ChatGPT — Promotional link to the creator's paid training, used as an example in the 'genius of evil' test.
  • Ressources IA gratuites — Free AI resources including prompts, mentioned as a source for the test prompts.
  • Vidéo de référence (cfRYp0nItZ8) — Referenced video, likely a previous test or tutorial, used for context.

Concurring Sources

  • OpenAI GPT-4.5 System Card — Official documentation supporting the model's capabilities and benchmarks.

External References

Contribution & Novelties

The video offers a practical, user-oriented evaluation of GPT-4.5, highlighting its improvements in natural interaction, conciseness, and role-playing adherence. It provides concrete examples that are useful for users, but the novelty is limited as it does not introduce new technical insights beyond the official documentation. The comparisons with other models are illustrative but not exhaustive.

Pour aller plus loin :

  • GPT-4.5 System Card — Official technical details on GPT-4.5’s training and capabilities.
  • Chain-of-thought prompting — Concept of step-by-step reasoning, relevant to GPT-4.5’s reasoning improvements.
  • Multilingual language models — Background on multilingual performance, relevant to GPT-4.5’s claimed improvements in French.

99 words

Radar Profile

The radar profile shows moderate scores across all dimensions, with quantity of information slightly higher than quality and technical level. This indicates a video that provides a good amount of practical information but lacks deep technical rigor and systematic evaluation.

Reliability 6/10

💬 Très positif. Sur les 30 commentaires analysés, les spectateurs expriment une forte appréciation pour la réactivité et la qualité pédagogique de la vidéo, avec de nombreux remerciements et encouragements.