DEV Community

달리기
달리기

Posted on

I pitted Claude and Gemini against each other while building a crypto bot. Here's what I found.

While building a crypto notification bot, I developed a weird habit: copying the same question into both Claude and Gemini, then comparing the answers.

My journal at the time: "Pasting the same questions into Gemini and Claude and watching them disagree. This is more fun than I expected."


Round 1: Setup time estimate

Asked both how long Make.com automation setup would take for a beginner.

  • Gemini: "20 minutes"
  • Claude: "That's quite optimistic for a first-timer. Expect 1-2 hours. Also, the free plan has a 1,000 operations/month limit — worth knowing upfront."

Tried it myself. Claude was right.


Round 2: Code review

  • Claude: "6.5/10. It works, but this code fails silently." (followed by a bug list)
  • Gemini: "100 out of 100... but here are 5 things to fix."

Gemini's praise felt good at first. After the third time, I stopped trusting it.


Round 3: Error fixing

Had an error related to Gemini's API. Asked Claude — didn't work. Asked Gemini directly — solved immediately.

Turns out, ask the tool about itself.


The moment that stuck

Asked Claude: "What if I charge people for bot signals?"

Claude: "Companies with billions in funding already do this. Selling investment signals for money can also be a legal issue. Honestly, the only realistic angle here is selling the process of building this — not the signals."

Didn't want to hear it. Wasn't wrong.

Journal after that: "There's this AI called Claude. More realistic than Gemini. Ended up buying it."


Takeaway

It's not about which one is better. It's about knowing where to use each.

  • Claude: uncomfortable truths, accurate bug spotting, realistic advice
  • Gemini: fast drafts, knows its own API best, friendly step-by-step

Using both at the same time showed me things I couldn't see using either alone.

Top comments (0)