Gemini Animate | AI Wisdom

← Back to topic

🎞 ANIMATE · Mode 3 of 5 · +100 XP— Google Gemini

📋Config

⚙️Generate

🔄Chat

✅Multimodal

🎯 Next: Test →

Step 1 / 4

Speed

What's happening

Configure Model

GenerativeModel with system_instruction and generation_config — one-time setup

💡 Analogy

Like configuring a customer service bot once at startup — same settings for all customers.

Configure Gemini with model name, system instructions, and generation parameters. GenerativeModel is reused across calls — efficient for high-volume use.

Console Output

> model: gemini-1.5-pro
> Context: 1M tokens (longest in industry)
> JSON mode: enabled
> system_instruction: coding assistant

●gemini.py

Configure ModelL6–14

1import google.generativeai as genai

2import os

4genai.configure(api_key=os.environ["GOOGLE_API_KEY"])

6# Gemini 1.5 Pro — 1M token context window

7model = genai.GenerativeModel(

8 model_name="gemini-1.5-pro",

9 system_instruction="You are a helpful coding assistant.",

10 generation_config=genai.types.GenerationConfig(

11 temperature=0.7,

12 max_output_tokens=1024,

13 response_mime_type="application/json", # JSON mode

14 ),

15)

17# Simple text generation

18response = model.generate_content("List top 3 Python best practices as JSON")

19print(response.text)

21# Multi-turn conversation (chat session)

22chat = model.start_chat(history=[])

23response1 = chat.send_message("My codebase uses FastAPI. What ORM should I use?")

24print(response1.text)

25response2 = chat.send_message("How do I set it up?") # knows context

26print(response2.text)

28# Multimodal: pass file or bytes

29import PIL.Image

30image = PIL.Image.open("architecture.png")

31vision_response = model.generate_content([

32 "Describe this system architecture",

33 image, # PIL Image directly supported

34])

35print(vision_response.text)

35 lines

Steps · 4

What's happening

Configure Model

GenerativeModel with system_instruction and generation_config — one-time setup

💡 Analogy

Like configuring a customer service bot once at startup — same settings for all customers.

Configure Gemini with model name, system instructions, and generation parameters. GenerativeModel is reused across calls — efficient for high-volume use.

Console Output

> model: gemini-1.5-pro
> Context: 1M tokens (longest in industry)
> JSON mode: enabled
> system_instruction: coding assistant

Was this page helpful?

Discussion

← Back to topic

🎞 ANIMATE · Mode 3 of 5 · +100 XP— Google Gemini

📋Config

⚙️Generate

🔄Chat

✅Multimodal

🎯 Next: Test →

Step 1 / 4

Speed

What's happening

Configure Model

GenerativeModel with system_instruction and generation_config — one-time setup

💡 Analogy

Like configuring a customer service bot once at startup — same settings for all customers.

Configure Gemini with model name, system instructions, and generation parameters. GenerativeModel is reused across calls — efficient for high-volume use.

Console Output

> model: gemini-1.5-pro
> Context: 1M tokens (longest in industry)
> JSON mode: enabled
> system_instruction: coding assistant

●gemini.py

Configure ModelL6–14

1import google.generativeai as genai

2import os

4genai.configure(api_key=os.environ["GOOGLE_API_KEY"])

6# Gemini 1.5 Pro — 1M token context window

7model = genai.GenerativeModel(

8 model_name="gemini-1.5-pro",

9 system_instruction="You are a helpful coding assistant.",

10 generation_config=genai.types.GenerationConfig(

11 temperature=0.7,

12 max_output_tokens=1024,

13 response_mime_type="application/json", # JSON mode

14 ),

15)

17# Simple text generation

18response = model.generate_content("List top 3 Python best practices as JSON")

19print(response.text)

21# Multi-turn conversation (chat session)

22chat = model.start_chat(history=[])

23response1 = chat.send_message("My codebase uses FastAPI. What ORM should I use?")

24print(response1.text)

25response2 = chat.send_message("How do I set it up?") # knows context

26print(response2.text)

28# Multimodal: pass file or bytes

29import PIL.Image

30image = PIL.Image.open("architecture.png")

31vision_response = model.generate_content([

32 "Describe this system architecture",

33 image, # PIL Image directly supported

34])

35print(vision_response.text)

35 lines

Steps · 4

What's happening

Configure Model

GenerativeModel with system_instruction and generation_config — one-time setup

💡 Analogy

Like configuring a customer service bot once at startup — same settings for all customers.

Configure Gemini with model name, system instructions, and generation parameters. GenerativeModel is reused across calls — efficient for high-volume use.

Console Output

> model: gemini-1.5-pro
> Context: 1M tokens (longest in industry)
> JSON mode: enabled
> system_instruction: coding assistant