I have noticed recently that Claude Code often sounds condescending, especially when explaining things. I feel being treated like child, gives me an impression of watching a Sesame Street episode, instead of working with a tool.Just today I ran Claude Code with Sonnet 5 and, to compare, OpenCode with a local Qwen 3.6, and asked both to explain me a pull request, same prompt, same CLAUDE.md file. Here are some excerpts:Sonnet title: Teaching the program to speak JSON Qwen title: JSON deserialization support for AMQP consumers Sonnet s description of a larger context: Rather than one giant program doing everything, the work is split across small applications that talk to each other through RabbitMQ .Qwen: This is a Python library that provides shared infrastructure for AMQP data processing applications: applications extend MessageConsumer class, implement process_one_message(), and the framework handles the rest, consuming messages from RabbitMQ, calling the application logic, and publishing results. Like, from the Sonnet s description alone I wouldn t even know what this code does for real!And the metaphors by Claude... From bytes to DataFrames, and back , ugh... Just call it desirialization, come on!And then later, Claude Code spent a paragraph explaining me how a lookup table is better than an `if .. else` ladder, duh... While Qwen actually focused directly on the nature of the change, not minute the implementation details.OK, I admit it Sonnet produced a super nice HTML report, Qwen s output is more reminiscent of a man page in HTML format. But... this kind of goes the same way, it just creates a impression of being professional.Did you notice anything like this? Is this new or did I just miss it before? Do you remediate this somehow?