I Tested GPT-5.5 vs Opus 4.7 on 8 Real Tasks. Here's Who Won.

Value of the Information & Strength of the Argument

The video provides a practical, hands-on evaluation of two leading AI models, which is valuable for users deciding which model to use for specific tasks. The methodology is clear: identical prompts, single-shot generation, and a third-party judge (Gemini) for some tasks. The creator’s commentary is insightful, highlighting specific strengths and weaknesses, such as Opus 4.7’s superior design taste and interactive elements, and GPT-5.5’s better data handling and dashboard organization. However, the evaluation is subjective, relying on the creator’s personal preferences and visual inspection rather than quantitative metrics. The use of Gemini as a judge is inconsistent, as it sometimes fails to render outputs correctly, leading the creator to override its decisions. Overall, the argumentation is coherent and well-structured, but the lack of rigorous testing criteria limits its scientific value.

Scientific Rigor, Source Quality, Title Accuracy

The video is a practical demonstration rather than a scientific study, so it does not cite academic sources. The creator provides a PDF with all prompts and code, which is a useful resource for reproducibility. The title accurately reflects the content, and the video’s structure is clear with chapters. The creator does not engage with external literature or benchmarks, relying solely on his own testing. The quality of sources is limited to the creator’s own observations and the outputs of the models. The video’s strength lies in its practical, real-world tasks, which are more relevant than synthetic benchmarks. However, the lack of rigorous methodology and reliance on subjective judgment reduce its scientific rigor.

The title accurately reflects the content: a head-to-head test of GPT-5.5 and Opus 4.7 on eight tasks, with a clear winner declared.

6/10

The video is a hands-on comparative test of two AI models, with clear methodology (same prompts, single-shot, third-party judge). However, the evaluation is subjective and lacks rigorous quantitative metrics, and the judge (Gemini) is not consistently reliable.

• 00:00Intro
• 00:38Test 1: Landing Page (with strict brand brief)
• 05:52Test 2: Landing Page (no design direction)
• 10:19Test 3: Email Client (Superhuman / Gmail clone)
• 12:49Test 4: E-commerce Dashboard (Stripe-style)
• 14:30Test 5: 3D Animal Cell (Three.js)
• 16:50Test 6: 2D Asteroid Game
• 18:26Real-World Office Tasks (intro)
• 19:33Test 7: Marketing Performance Deck (VP of Growth)
• 22:26Test 8: Finance Board Pack (CFO / Audit Committee)
• 25:48Final Tally + Verdict

• PDF of all tests (with prompts and code) — Provided by the creator for reference, containing all prompts and code used in the tests.
• AI bootcamp (persimmons.studio) — Mentioned in the description as a paid bootcamp, not directly related to the video's content.
• AI For Mortals newsletter — Mentioned in the description as a newsletter, not directly related to the video's content.

• Claude 4.7 (Anthropic) — Official page for Claude, which includes Opus 4.7, the model tested.
• GPT-5.5 (OpenAI) — Official page for GPT-5, which may include GPT-5.5 details.