🚨NOTICIAS IA🚨: La GRAN Cagada de GPT-5 🤬❌

🚨NOTICIAS IA🚨: La GRAN Cagada de GPT-5 🤬❌

by John Hernández

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQFGenerative artificial intelligence
🎙 John Hernández 👥 725K 📅 August 11, 2025 ⏱ 30 min 👁 167K 📄 news review 🧭 2026-08-03
Available in: English (current) Français

Keywords

GPT-5routerlimitsbenchmarksuser complaints

Summary

This video by John Hernández reviews the latest AI news, focusing on the controversial launch of OpenAI’s GPT-5. The creator explains that while GPT-5 performs well on benchmarks, especially in coding and reasoning, users are complaining about poor responses due to a faulty router that often sends queries to the weaker ‘minimal’ model. He details the new usage limits for free and Plus accounts, noting that forcing ’thinking’ mode can improve results but consumes separate quotas. OpenAI has responded by tweaking the router and temporarily increasing limits, but the core issue persists. The video also covers other news: Google’s Genie 3 world model, Anthropic’s Claude 4.1, ElevenLabs’ music generation, and an AI discovering new physics. Hernández provides practical tips for using GPT-5 effectively, such as adding prompts to encourage deep thinking. He also addresses the emotional attachment some users have to GPT-4o and explains how to access legacy models. The video includes a sponsor segment for Hostinger and promotes his AI community. Overall, it offers a balanced, critical analysis of GPT-5’s launch, highlighting both its strengths and the user experience problems.

181 words

Critical Evaluation

The video provides a comprehensive and critical analysis of the GPT-5 launch, effectively balancing technical details with user feedback. Hernández demonstrates a solid understanding of the model’s architecture, explaining the hybrid nature of GPT-5 and how the router’s malfunction leads to subpar responses. He supports his claims with references to benchmarks like Artificial Analysis and METR, which adds credibility. The discussion of usage limits is thorough, with specific numbers for different account tiers, and he offers practical advice for mitigating the router issues, such as prompting the model to think deeply. The emotional backlash from users is presented with empathy, acknowledging the human tendency to form attachments to AI, which is a relevant sociological observation. However, the video is not without flaws. The title’s sensationalism (‘La GRAN Cagada’) may overstate the severity of the issues, though the content does justify the criticism. The reliance on user comments and anecdotal evidence could introduce bias, but Hernández does attempt to contextualize these by noting that the model itself is strong on benchmarks. The sponsor segment is clearly disclosed and does not detract from the content’s integrity. The video’s strength lies in its timeliness and practical value for users navigating the new model. The adéquation between title and content is good, as the video indeed focuses on the problems and backlash. Overall, the video is a valuable resource for those interested in AI developments, offering a balanced perspective that combines technical analysis with user experience.

242 words

Title / Content Match

The title accurately reflects the video's focus on the problems and backlash surrounding GPT-5's launch, using sensational language that matches the content.

Quality & Reliability

7/10

The video provides a detailed and timely review of GPT-5's launch issues, supported by benchmark data and official sources. The creator demonstrates technical knowledge and offers practical advice. However, the content is opinionated and relies on user feedback, which introduces subjectivity. The presence of a sponsor segment is disclosed but does not affect the score.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

Dissenting Sources

External References

Contribution & Novelties

The video offers a timely and critical analysis of GPT-5’s launch, highlighting the router issue and its impact on user experience. It provides practical advice for users to mitigate the problems, such as prompting the model to think deeply. The coverage of user emotional attachment to GPT-4o adds a unique sociological perspective. The video also aggregates other AI news, giving a comprehensive overview of the week’s developments.

Pour aller plus loin :

  • Artificial Analysis — Independent benchmark aggregator for AI models, useful for comparing GPT-5 with others.
  • METR — Research organization measuring AI capabilities, referenced in the video for long-task benchmarks.
  • OpenAI Help Center — Official documentation for GPT-5 and other models, providing detailed information on features and limits.
  • Google DeepMind Blog — Source for Genie 3 and other AI research updates.
  • Anthropic News — Official announcements from Anthropic, including Claude 4.1.

142 words

Radar Profile

The radar profile shows a balanced performance across all dimensions, with slightly higher scores in quantity of information and technical level, reflecting the video's comprehensive coverage and technical depth. The lower score in reliability is due to the reliance on user feedback and the creator's subjective opinions.

Reliability 7/10

💬 Négatif. Sur les 30 commentaires analysés, la majorité exprime des plaintes concernant la dégradation de l'expérience avec GPT-5, notamment la lenteur, les limites réduites et la perte de personnalisation, bien que certains reconnaissent des améliorations en programmation.