Key Tips for Using Google Gemini

Using settings effectively is crucial for getting the most out of any AI, including Gemini. However, Gemini has detailed configuration options, and setting them incorrectly can lead to wasted time and unnecessary usage quota consumption without making progress.

Here is a breakdown of Gemini's mode settings. Selecting the right options is key to making the most of your usage limit. Many of you likely subscribe to the Pro plan (Google One AI Premium, around 2,900 yen/month). However, if you fully utilize Gemini 3, you can hit the limits quickly. Below is a summary of key optimization tips.

＜Basic Tips＞

For high-performance models like Gemini 3 Pro, there is no weekly usage limit. Instead, a 'gradual' reset occurs every 5 hours. Unlike other AIs where limits are reset all at once after a set window, Gemini's quota recovers incrementally. It usually takes about a day for a full reset. Note that running heavy processes repeatedly can penalize you, extending the reset window to several days.
Models like Flash and Flash-Lite can be used extensively. For most of your routine tasks in Gemini, these lightweight modes will likely make up the bulk of your work.

※Image and music generation have separate limits.

＜Modes＞

Mode Type	Overview
Default	Standard chat mode. Features: Polite, neutral, and balanced responses. Behavior: Understands user intent while avoiding excessive decoration, providing factual information. Uses: Everyday questions, research, general email writing, etc.
Auto Edit	An 'editor' mode specialized in proofreading and refining text. Features: Focuses on correcting typos, grammar, and logical structure of the input text. Behavior: Minimizes 'chatter' with the user, outputting the revised text immediately. It may include a brief explanation of why the edits were made. Uses: Proofreading essays, polishing business emails, and debugging code.
YOLO	A fast and decisive mode. Features: True to the name 'You Only Live Once,' it reaches conclusions without hesitation. Behavior: Eliminates introductory filler or ambiguous phrases like 'it depends,' presenting the single 'best choice' according to Gemini. The tone may be bold or humorous. Uses: Seeking a push for decision-making, or looking for sharp opinions during brainstorming.
Plan	A 'designer' mode focused on structure and procedure. Features: Creates an 'execution plan' or 'outline' before reaching a final conclusion. Behavior: Breaks down complex tasks step-by-step ($Step 1, Step 2 \dots$), showing the big picture before entering into details. Uses: Writing long articles, planning trips, programming design, and creating learning roadmaps. As file editing is automatically disabled, it is not suitable for project execution.

＜Model Generation Features＞

Model Name

Gemini 3 (Latest / Next Gen)

Features: The latest architecture as of 2026. Significantly outperforms 2.5 in reasoning, code generation, and long-context understanding.

Uses: Complex logic puzzles, advanced math, professional programming, and analyzing very long documents.

Gemini 2.5 (Stable)

Features: A highly reliable and balanced model. It was the flagship before Gemini 3 and remains powerful enough for many use cases.

Uses: General writing, everyday questions, standard data analysis, etc.

＜Model Types＞

Type	Intelligence (Reasoning)	Speed / Cost	Main Uses
Pro	Strongest	Normal	Complex problem solving, creative writing, advanced reasoning
Flash	High	Fast	Real-time dialogue, summarizing large text volumes, fast responses
Flash-Lite	Standard	Super Fast	Simple tasks, high-volume API processing, lightweight chat

＜Plan List＞

Service Plan	Price	Models Used	Description
Free	Free	Gemini	Coding not supported
AI Plus 200MB	1,200 yen /month	Chat: Gemini Images: Nano Banana Video: Veo 3.1 Fast Music: Lyria 3 Extended	Coding not supported
AI Pro 2TB	2,900 yen /month	Chat: Gemini Images: Nano Banana Video: Veo 3.1 Fast Music: Lyria 3 Premium
AI Ultra 30TB	36,400 yen /month	Chat: Gemini Images: Nano Banana Video: Veo 3.1 Music: Lyria 3 Ultimate

＜Gemini Folders＞

Gemini stores information and shared context (for the same user) gathered across threads (a sequence of conversation from start to finish) in two folders to ensure smooth interactions in subsequent threads.

.gemini directly under the home directory (~/.gemini)
Role: Stores global configurations and data shared across the entire system, such as API keys, login information, past chat history, and temporary AI work files like the ones used here.
Creation Timing: Automatically created when the Gemini CLI is installed and configured.
.gemini directly under the project folder (./.gemini)
Role: Created manually when you want to define project-specific rules, such as coding guidelines or specific instructions for the AI.
Priority: If this folder exists in the project root, the AI prioritizes project-specific settings over the global home directory settings.