ChatGPT, Claude, Grok and Gemini: Peter Yang's Picks for 10 Everyday AI Tasks

Peter Yang compares ChatGPT, Claude, Grok and Gemini across ten practical uses. His overall choice is ChatGPT, but he stresses that the answer depends on the job. These are his impressions from examples and existing workflows, not a controlled benchmark.

For design, Yang asks each model to create a recipe application. ChatGPT produces a basic interface without recipe pictures; Claude's initial version works but looks generic. Claude Design asks useful questions, though its result also lacks pictures. Grok produces the visual design he likes best from this prompt, while he finds alignment problems in Gemini's result.

Yang still favors Claude for the broader design category. He shows a slide deck generated using a custom skill and brand assets, and mentions websites and animated videos as other strengths. Those existing instructions and assets matter when interpreting the comparison: the tools are not all starting with identical context.

For everyday conversation he prefers ChatGPT's personality and finds Claude less natural for his needs. This is a personal preference, not an assessment of factual reliability; he does not give Grok and Gemini the same full demonstration in this section.

For writing and editing, Yang prefers ChatGPT. He values its ability to work with his voice and produce useful revisions, although the quality of any draft still depends on the prompt, context and human review.

For planning and difficult decisions, he chooses Claude. In one business-planning example, it suggested an operations hire that he says ChatGPT missed. One recommendation is not proof that a model will make better decisions in every situation; the surrounding business context matters.

For coding, Yang's pick is ChatGPT. He notes that he is not an engineer and does not inspect the generated code himself. A developer friend has a different view, saying GPT and Claude can overengineer and preferring Grok. The disagreement illustrates why code review and testing matter more than a single ranking.

Yang considers browser and computer use an underrated category and favors ChatGPT for carrying out tasks across websites. A useful demonstration still needs a person to check the action and result, especially when an agent encounters sensitive data or consequential operations.

For voice conversations, Yang prefers ChatGPT's experience. His judgment reflects the interfaces available when the video was recorded, and can change as products and plans change.

In image generation he favors ChatGPT's adherence to his prompts. He acknowledges that it already has access to brand and conversational context that the other tools do not, making this an uneven comparison rather than a clean test of image models alone.

For video generation, he picks Gemini based on an example of a Japanese family commercial. That single output is an illustration of the type of result he liked, not a broad evaluation of video quality or a guarantee that the same prompt will work elsewhere.

For personal agents, Yang again chooses ChatGPT and describes it as useful for ongoing work. He ends with a scorecard showing ChatGPT as his overall winner, while retaining Claude as his design and planning choice and noting Grok's strong showing in app design.

The useful takeaway is to compare tools on your own recurring work: supply comparable context, inspect the output and account for custom skills or prior memory. Yang's examples are a snapshot from August 2026, not a permanent ranking of fast-changing products.