I Made Opus 4.8 and Fable 5 Build the Same App (RAW RESULTS)

I Made Opus 4.8 and Fable 5 Build the Same App (RAW RESULTS)

by Pat Simmons

Applied Sciences & Engineering AI — Models & Research UYQArtificial intelligence (AI)UYQMMachine learning
🎙 Pat Simmons 👥 24K 📅 June 11, 2026 ⏱ 21 min 👁 933K 📄 original study 🧭 2026-09-07
Available in: English (current) Français

Keywords

AI codingmodel comparisone-commerce3D museumgame development

Summary

In this video, Pat Simmons conducts a head-to-head comparison between Claude Opus 4.8 and Anthropic’s new Fable 5 model by giving them the same prompts to build three distinct applications: an e-commerce store, an interactive 3D art museum, and an Age of Empires-style strategy game. Each build is performed in a single one-shot output with no revisions, and the results are deployed live. The video documents the process, including token usage, cost estimates, and build times. For the e-commerce store, Fable 5 produced a more polished and user-friendly interface with better image generation and filtering, while Opus 4.8 had minor UX issues. The 3D art museum was a more complex challenge; Fable 5 successfully created an immersive, navigable 3D gallery with smooth interactions, whereas Opus 4.8’s version had broken navigation and could not enter the galleries. The final build, an Age of Empires clone, was the most demanding: Opus 4.8 produced a non-functional, broken game, while Fable 5 delivered a fully playable 3D RTS game with impressive graphics and mechanics. Throughout the video, Simmons highlights Fable 5’s superior speed, token efficiency, and design intuition, despite its higher usage-based cost. The video concludes that Fable 5 is the clear winner across all three builds, showcasing a significant leap in AI coding capabilities.

211 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides valuable empirical data on the performance of two leading AI models in real-world coding tasks. The methodology is straightforward: identical prompts, one-shot outputs, and live deployment, which adds credibility to the comparison. The author includes quantitative metrics like token usage and cost, offering practical insights for developers. However, the evaluation is largely subjective, relying on visual inspection and personal preference rather than objective benchmarks. The argumentation is persuasive but could be strengthened by more rigorous testing, such as multiple runs or blind evaluation. The video effectively demonstrates the capabilities of Fable 5, but the lack of control for potential biases (e.g., prompt wording, model updates) limits the generalizability of the findings.

Scientific Rigor, Source Quality, Title Accuracy

The video does not cite external sources, but it references the models’ capabilities and Anthropic’s announcements. The title accurately reflects the content, and the video’s claims are based on direct experimentation. The author’s expertise is evident, but the lack of peer-reviewed sources or independent verification reduces the scientific rigor. The video is well-structured with clear chapters, and the author transparently shares his process and limitations. The comments section shows high engagement and positive reception, with viewers expressing amazement at Fable 5’s performance, though some note the cost implications. Overall, the video is a credible demonstration but not a formal scientific study.

230 words

Title / Content Match

The title accurately reflects the content: a direct comparison of two AI models building the same applications.

Quality & Reliability

7/10

The video presents a hands-on comparative test of two AI models with clear methodology, but lacks formal controls and relies on subjective evaluation.

Chapters

Cited Sources

Concurring Sources

  • Anthropic's Fable 5 announcement — The video references Anthropic's claims about Fable 5's capabilities, which are consistent with the observed performance.

Dissenting Sources

  • Community reports on Fable 5 speed — Some comments and external discussions suggest Fable 5 may be slower than Opus 4.8 in certain tasks, contradicting the video's findings.

Contribution & Novelties

The video offers a novel, hands-on comparison of two cutting-edge AI models in complex, real-world coding tasks, providing insights into their practical strengths and weaknesses. It highlights Fable 5’s superior performance in terms of speed, token efficiency, and design quality, which is valuable for developers considering model adoption.

Pour aller plus loin :

  • Claude models overview — Official documentation on Claude models, including capabilities and pricing.
  • Three.js — The JavaScript library used for 3D graphics in the museum and game builds.
  • Wikimedia Commons API — The API used to fetch art images for the museum.

95 words

Radar Profile

The radar chart shows high scores in information quantity and technical level, reflecting the detailed and hands-on nature of the video. The quality and reliability scores are moderate, indicating a well-executed but subjective comparison.

Reliability 7/10

💬 Très positif. Sur les 30 commentaires analysés, la grande majorité exprime un enthousiasme marqué pour les capacités de Fable 5, avec des réactions d'étonnement et d'admiration, bien que certains soulignent le coût élevé et la rapidité du progrès technologique.