Pular para o conteúdo principal

Chat with AI - Overview

Chat is the core feature of AI Studio Plus, giving you access to the world's most advanced language models in one unified interface.

What is AI Chat?​

AI Chat lets you have natural conversations with state-of-the-art language models. Whether you need help writing, coding, analyzing data, brainstorming ideas, or learning new concepts, AI chat is your intelligent assistant.

Available AI Models​

Access multiple leading AI models from a single platform:

OpenAI Models​

GPT-4o (Latest & Most Capable)

  • Best for: Complex reasoning, analysis, multi-step tasks
  • Strengths: Problem-solving, code generation, detailed explanations
  • Cost: 0.5-1.0 credits per message
  • Context: 128K tokens

GPT-4 Turbo

  • Best for: Advanced conversations, creative writing
  • Strengths: Consistency, accuracy, nuanced understanding
  • Cost: 0.3-0.7 credits per message
  • Context: 128K tokens

GPT-3.5 Turbo

  • Best for: Quick tasks, simple queries, high-volume usage
  • Strengths: Fast, cost-effective, reliable
  • Cost: 0.1-0.2 credits per message
  • Context: 16K tokens

Anthropic Models​

Claude 3.5 Sonnet (Recommended)

  • Best for: Writing, creativity, analysis, coding
  • Strengths: Long context, nuanced responses, safety
  • Cost: 0.4-0.8 credits per message
  • Context: 200K tokens

Claude 3 Opus

  • Best for: Most demanding tasks, highest quality
  • Strengths: Exceptional reasoning, creativity, accuracy
  • Cost: 0.6-1.2 credits per message
  • Context: 200K tokens

Claude 3 Haiku

  • Best for: Fast responses, simple tasks
  • Strengths: Speed, efficiency, cost-effectiveness
  • Cost: 0.1-0.3 credits per message
  • Context: 200K tokens

Google Models​

Gemini 1.5 Pro

  • Best for: General tasks, multimodal inputs
  • Strengths: Long context, versatility, vision capabilities
  • Cost: 0.2-0.4 credits per message
  • Context: 1M tokens

Gemini 1.5 Flash

  • Best for: Fast, lightweight tasks
  • Strengths: Speed, efficiency
  • Cost: 0.1-0.2 credits per message
  • Context: 1M tokens

Open Source Models​

Llama 3 70B

  • Best for: Privacy-focused tasks, experimentation
  • Strengths: Open source, capable, cost-effective
  • Cost: 0.2-0.4 credits per message
  • Context: 8K tokens

Mixtral 8x7B

  • Best for: Multilingual tasks, efficient inference
  • Strengths: Strong reasoning, code generation
  • Cost: 0.15-0.3 credits per message
  • Context: 32K tokens

Key Features​

1. Conversation History​

  • Persistent Conversations: All chats are automatically saved
  • Context Awareness: AI remembers previous messages in the conversation
  • Search & Filter: Find past conversations quickly
  • Export: Download conversations as TXT, MD, or PDF

2. File Attachments​

Upload files for AI analysis:

  • Documents: PDF, DOCX, TXT, MD
  • Images: JPG, PNG (with vision-capable models)
  • Code: Any programming language
  • Data: CSV, JSON, XML
Vision Models

GPT-4o, GPT-4 Turbo, Claude 3 models, and Gemini support image analysis. Upload screenshots, diagrams, charts, or photos for AI to analyze.

3. Code Execution​

Some models can execute code:

  • Python: Run Python code snippets
  • JavaScript: Execute JS in sandbox
  • Data Analysis: Process CSV/JSON data
  • Visualizations: Generate charts and graphs

Enable web search for current information:

  • Real-time Data: Get up-to-date information
  • Fact Checking: Verify claims with sources
  • Research: Deep dive into topics
  • Citations: See sources used
Coming Soon

Web search is currently in beta. Enable in Settings > Experimental Features.

5. Custom Instructions​

Set persistent preferences for all conversations:

  • Writing Style: "Always write in a professional tone"
  • Response Format: "Use bullet points for lists"
  • Context: "I'm a software developer working in Python"
  • Constraints: "Keep responses under 200 words"

Common Use Cases​

Writing & Content Creation​

  • Blog posts and articles
  • Social media content
  • Email drafts
  • Product descriptions
  • Creative stories
  • Copywriting and ads
  • Scripts and screenplays

Coding & Development​

  • Code generation
  • Debugging assistance
  • Code review and optimization
  • Documentation writing
  • Architecture design
  • Database queries
  • API integration help

Business & Productivity​

  • Meeting summaries
  • Report writing
  • Data analysis
  • Business plans
  • Presentations
  • Email responses
  • Research and insights

Learning & Education​

  • Explain complex concepts
  • Homework help
  • Study guides
  • Language practice
  • Historical research
  • Scientific explanations
  • Math problem solving

Creative & Brainstorming​

  • Idea generation
  • Product naming
  • Campaign concepts
  • Story development
  • Character creation
  • Marketing strategies
  • Problem-solving approaches

Getting Started with Chat​

Basic Workflow​

  1. Select a Model - Choose based on your task and budget
  2. Type Your Message - Be clear and specific
  3. Review Response - Read the AI's answer
  4. Follow Up - Ask clarifying questions or request changes
  5. Save or Export - Keep important conversations

Example Conversation Flow​

You: I need to write a professional email to my boss requesting time off.

AI: I'd be happy to help you draft that email. To make it perfect for your
situation, could you tell me:
1. How much time off do you need?
2. What dates are you requesting?
3. The reason (if you're comfortable sharing)?
4. How formal is your workplace culture?

You: I need 5 days off from June 15-19 for a family vacation. We have a
pretty casual workplace.

AI: Here's a draft email for you:

Subject: Time Off Request - June 15-19

Hi [Boss's Name],

I wanted to request time off from June 15-19 (5 days) for a family vacation
we've been planning.

I'll make sure all my current projects are on track before I leave, and I'm
happy to brief [Colleague's Name] on anything that might need attention
while I'm away.

Please let me know if these dates work or if you need any additional
information.

Thanks!
[Your Name]

Would you like me to adjust the tone or add anything?

You: Can you make it a bit more formal?

AI: [Provides more formal version]

Tips for Effective Chat​

1. Be Specific​

❌ "Tell me about Python" ✅ "Explain how list comprehensions work in Python with examples"

2. Provide Context​

❌ "Fix this code: [code]" ✅ "This Python function should filter even numbers but returns an error. Can you identify and fix the issue? [code]"

3. Break Down Complex Tasks​

Instead of: "Create a complete business plan" Try:

  1. "Help me outline a business plan structure"
  2. "Write the executive summary section"
  3. "Develop the market analysis section"
  4. etc.

4. Use Follow-Up Prompts​

  • "Make it more concise"
  • "Add more details about X"
  • "Rewrite in a friendly tone"
  • "Give me 3 variations"
  • "Explain that differently"

5. Iterate and Refine​

Don't expect perfection on the first try. Use the conversation to refine:

  • Request changes
  • Ask for alternatives
  • Provide feedback
  • Build on responses

Advanced Features​

Temperature Control​

Adjust creativity vs. consistency:

  • 0.0-0.3: Deterministic, focused, consistent (good for factual tasks)
  • 0.4-0.7: Balanced (default)
  • 0.8-1.0: Creative, varied, exploratory (good for brainstorming)

System Prompts​

Define how the AI should behave:

"You are an expert Python developer. Provide code examples for every
explanation. Prioritize best practices and efficiency."

Response Length​

Control output length:

  • Short: Quick, concise answers
  • Medium: Balanced detail (default)
  • Long: Comprehensive, detailed responses

Chat Templates​

Save common prompt structures:

  • Email templates
  • Code review format
  • Content outlines
  • Research queries

Best Practices​

Security & Privacy​

  • ✅ Don't share sensitive personal information
  • ✅ Don't paste passwords or API keys
  • ✅ Don't include confidential business data
  • ✅ Review AI responses before using in production
  • ✅ Fact-check critical information

Ethical Usage​

  • ✅ Use AI to enhance your work, not replace critical thinking
  • ✅ Disclose AI assistance when required
  • ✅ Review and edit AI-generated content
  • ✅ Don't use for academic dishonesty
  • ✅ Respect copyright and attribution

Cost Optimization​

  • Use GPT-3.5 or Gemini Flash for simple tasks
  • Use GPT-4 or Claude for complex reasoning
  • Keep prompts concise to reduce token usage
  • End conversations when no longer needed
  • Use projects to organize and reuse conversations

Limitations​

AI chat models have limitations:

Knowledge Cutoff​

  • Models have a knowledge cutoff date
  • May not know recent events
  • Enable web search for current information

Hallucinations​

  • AI can generate plausible but incorrect information
  • Always verify critical facts
  • Use multiple sources for important decisions

Context Limits​

  • Conversations have token limits
  • Very long chats may lose early context
  • Start new conversation for unrelated topics

Reasoning Limits​

  • AI can make logical errors
  • May struggle with math (use code execution)
  • Verify complex reasoning independently

Next Steps​

Ready to start chatting?

  1. Choose Your Model - Learn about each AI model in detail
  2. Master Prompts - Write effective prompts
  3. Manage Conversations - Organize your chats
  4. Try Your First Chat - Follow step-by-step tutorial

FAQs​

Q: Which model should I use? A: For most tasks, start with GPT-4o or Claude 3.5 Sonnet. Use GPT-3.5 for simple tasks to save credits.

Q: Can the AI access the internet? A: With web search enabled (beta), yes. Otherwise, it relies on training data.

Q: Is my conversation private? A: Yes. Conversations are encrypted and not used to train models without consent.

Q: Can I delete conversations? A: Yes. Click the trash icon on any conversation to permanently delete it.

Q: How long is conversation history kept? A: Indefinitely, unless you delete it. Export important conversations for backup.

Ready to chat? Open AI Studio Plus