WritingClaude wins
Claude remains the writing benchmark. Gemini is solid but more generic under pressure.
CodingClaude wins
Claude Code and agentic workflows give Claude the professional edge. Gemini 3.8 Flash is explicitly aimed at long-horizon engineering and has closed ground, but it is less battle-tested in serious pipelines.
Context sizeTie
A million tokens each. Neither has an advantage here any more.
PriceGemini wins
The free tier alone settles it for students and casual use. By API it is not close: $0.75 per million input tokens on Gemini 3.8 Flash against $5 on Opus 5 and $10 on Fable 5.1.
ResearchGemini wins
Search grounding with citations, NotebookLM integration and Workspace access make Gemini the stronger study partner.
On paper