Perplexity API Guide for Developers — 2026

# Get your API key at perplexity.ai/settings/api
export PERPLEXITY_API_KEY="pplx-..."
 
# Use the openai Python package — no additional library needed
pip install openai

from openai import OpenAI
 
client = OpenAI(
    api_key="pplx-...",  # Or use PERPLEXITY_API_KEY env variable
    base_url="https://api.perplexity.ai",
)
 
def search_and_answer(question: str, model: str = "llama-3.1-sonar-large-128k-online") -> str:
    """
    Ask a question with real-time web search.
    Returns the answer text.
    """
    response = client.chat.completions.create(
        model=model,
        messages=[
            {
                "role": "system",
                "content": (
                    "You are a technical research assistant. "
                    "Provide accurate, current information with specific details. "
                    "Mention version numbers and dates when relevant."
                ),
            },
            {"role": "user", "content": question},
        ],
        max_tokens=1024,
        temperature=0.2,
    )
 
    return response.choices[0].message.content
 
answer = search_and_answer(
    "What is the latest stable version of Rust and what changed in the most recent release?"
)
print(answer)

from openai import OpenAI
 
client = OpenAI(
    api_key="pplx-...",
    base_url="https://api.perplexity.ai",
)
 
def search_with_citations(question: str) -> dict:
    """Return both the answer and its citations."""
    response = client.chat.completions.create(
        model="llama-3.1-sonar-large-128k-online",
        messages=[{"role": "user", "content": question}],
        max_tokens=1024,
        temperature=0.2,
    )
 
    answer = response.choices[0].message.content
 
    # Citations are in the response object under citations attribute
    citations = getattr(response, "citations", [])
 
    return {
        "answer": answer,
        "citations": citations,
        "model": response.model,
    }
 
result = search_with_citations(
    "What are the breaking changes in React 19?"
)
print(result["answer"])
print("\nSources:")
for url in result.get("citations", []):
    print(f"  - {url}")

from openai import OpenAI
import json
 
client = OpenAI(
    api_key="pplx-...",
    base_url="https://api.perplexity.ai",
)
 
def check_library_update(library_name: str, current_version: str) -> dict:
    """Check if a library has been updated and what changed."""
    question = (
        f"What is the latest stable version of {library_name}? "
        f"The version I'm currently using is {current_version}. "
        f"If there is a newer version, what are the key changes and any breaking changes?"
    )
 
    response = client.chat.completions.create(
        model="llama-3.1-sonar-large-128k-online",
        messages=[
            {
                "role": "system",
                "content": (
                    "Return your answer as a JSON object with fields: "
                    "latest_version, is_update_available, breaking_changes, "
                    "key_changes, upgrade_recommended. "
                    "For upgrade_recommended: true if major security or performance improvements, false otherwise."
                ),
            },
            {"role": "user", "content": question},
        ],
        max_tokens=512,
        temperature=0.0,
    )
 
    try:
        text = response.choices[0].message.content
        # Extract JSON from response
        start = text.find("{")
        end = text.rfind("}") + 1
        return json.loads(text[start:end])
    except Exception:
        return {"error": "Could not parse response", "raw": response.choices[0].message.content}
 
# Check multiple dependencies
dependencies = [
    ("fastapi", "0.103.0"),
    ("pydantic", "2.4.0"),
    ("httpx", "0.25.0"),
]
 
for lib, version in dependencies:
    result = check_library_update(lib, version)
    print(f"\n{lib} ({version}):")
    print(f"  Latest: {result.get('latest_version', 'unknown')}")
    print(f"  Update available: {result.get('is_update_available', 'unknown')}")
    if result.get("breaking_changes"):
        print(f"  Breaking changes: {result['breaking_changes']}")

from openai import OpenAI
 
client = OpenAI(
    api_key="pplx-...",
    base_url="https://api.perplexity.ai",
)
 
class ResearchSession:
    """A multi-turn research session using Perplexity."""
 
    def __init__(self, topic: str):
        self.messages = [
            {
                "role": "system",
                "content": (
                    f"You are a research assistant helping investigate: {topic}. "
                    "Use current web search for all answers. "
                    "Be specific with versions, dates, and sources."
                ),
            }
        ]
 
    def ask(self, question: str) -> str:
        self.messages.append({"role": "user", "content": question})
 
        response = client.chat.completions.create(
            model="llama-3.1-sonar-large-128k-online",
            messages=self.messages,
            max_tokens=1024,
            temperature=0.2,
        )
 
        answer = response.choices[0].message.content
        self.messages.append({"role": "assistant", "content": answer})
        return answer
 
# Research a technology decision
session = ResearchSession("Choosing between PostgreSQL and MongoDB in 2026")
print(session.ask("What are the current production adoption rates?"))
print(session.ask("What do recent benchmarks show for write performance?"))
print(session.ask("Are there any significant updates to either in the past 6 months?"))