Basic Usage
Python
JavaScript
cURL
Go
from openai import OpenAI
client = OpenAI(
base_url="https://ai.megallm.io/v1",
api_key="your-api-key"
)
response = client.chat.completions.create(
model="gpt-4",
messages=[
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "What's the weather like?"}
],
temperature=0.7,
max_tokens=150
)
print(response.choices[0].message.content)
Advanced Features
Message Roles
The API supports different message roles for conversation context:
| Role | Description | Example |
|---|
system | Sets behavior and context | ”You are a helpful assistant” |
user | User input/questions | ”What’s the capital of France?” |
assistant | AI responses | ”The capital of France is Paris” |
tool | Tool/function results | Function execution results |
Temperature Control
Adjust response creativity with the temperature parameter:
# More deterministic (0.0 - 0.3)
response = client.chat.completions.create(
model="gpt-4",
messages=[...],
temperature=0.2 # More focused, consistent responses
)
# Balanced (0.4 - 0.7)
response = client.chat.completions.create(
model="gpt-4",
messages=[...],
temperature=0.5 # Balanced creativity and coherence
)
# More creative (0.8 - 1.0)
response = client.chat.completions.create(
model="gpt-4",
messages=[...],
temperature=0.9 # More varied, creative responses
)
Multi-turn Conversations
Maintain context across multiple exchanges:
messages = [
{"role": "system", "content": "You are a math tutor."},
{"role": "user", "content": "What is 15 * 12?"},
{"role": "assistant", "content": "15 * 12 = 180"},
{"role": "user", "content": "Now divide that by 6"}
]
response = client.chat.completions.create(
model="gpt-4",
messages=messages
)
# Response understands "that" refers to 180
Vision Support
Process images alongside text (requires vision-capable models):
response = client.chat.completions.create(
model="gpt-4-vision",
messages=[
{
"role": "user",
"content": [
{"type": "text", "text": "What's in this image?"},
{
"type": "image_url",
"image_url": {
"url": "https://example.com/image.jpg",
"detail": "high" # "low", "high", or "auto"
}
}
]
}
],
max_tokens=300
)
Standard Response
{
"id": "chatcmpl-abc123",
"object": "chat.completion",
"created": 1677858242,
"model": "gpt-4",
"usage": {
"prompt_tokens": 13,
"completion_tokens": 17,
"total_tokens": 30
},
"choices": [
{
"message": {
"role": "assistant",
"content": "The capital of France is Paris."
},
"finish_reason": "stop",
"index": 0
}
]
}
Finish Reasons
| Reason | Description |
|---|
stop | Natural completion |
length | Hit max_tokens limit |
tool_calls | Model wants to call a function |
content_filter | Content was filtered |
Best Practices
System Messages: Always include a clear system message to set the AI’s behavior and context.
Token Optimization
# Count tokens before sending
import tiktoken
encoding = tiktoken.encoding_for_model("gpt-4")
tokens = encoding.encode("Your message here")
print(f"Token count: {len(tokens)}")
# Truncate if needed
if len(tokens) > 1000:
truncated = encoding.decode(tokens[:1000])
Error Handling
try:
response = client.chat.completions.create(...)
except openai.APIError as e:
print(f"API error: {e}")
except openai.RateLimitError as e:
print(f"Rate limit hit: {e}")
# Implement exponential backoff
except openai.APIConnectionError as e:
print(f"Connection error: {e}")
# Retry with backoff
Common Patterns
Summarization
response = client.chat.completions.create(
model="gpt-4",
messages=[
{"role": "system", "content": "You are a summarization expert."},
{"role": "user", "content": f"Summarize this text in 3 bullet points: {long_text}"}
],
temperature=0.3,
max_tokens=150
)
Classification
response = client.chat.completions.create(
model="gpt-4",
messages=[
{"role": "system", "content": "Classify the sentiment as positive, negative, or neutral."},
{"role": "user", "content": "I love this product! It works great."}
],
temperature=0.0,
max_tokens=10
)
Code Generation
response = client.chat.completions.create(
model="gpt-4",
messages=[
{"role": "system", "content": "You are a Python expert."},
{"role": "user", "content": "Write a function to calculate fibonacci numbers"}
],
temperature=0.2,
max_tokens=500
)
Rate Limiting
Implement proper rate limiting to avoid errors:
import time
from typing import Optional
class RateLimiter:
def __init__(self, requests_per_minute: int = 60):
self.requests_per_minute = requests_per_minute
self.request_times = []
def wait_if_needed(self):
now = time.time()
# Remove requests older than 1 minute
self.request_times = [t for t in self.request_times if now - t < 60]
if len(self.request_times) >= self.requests_per_minute:
sleep_time = 60 - (now - self.request_times[0])
if sleep_time > 0:
time.sleep(sleep_time)
self.request_times.append(now)
# Usage
limiter = RateLimiter(60)
for prompt in prompts:
limiter.wait_if_needed()
response = client.chat.completions.create(...)
Next Steps
