As a data analyst who relies heavily on AI for summarizing reports, I found the 7.0% hallucination rate for Gemini pretty impressive compared to others I’ve tested. I usually use a Python and R workflow with ChatGPT integrated for quick insights Terminal-Bench 2.0 scores