Ollama: OpenAI兼容API

from openai import OpenAI

# Before: OpenAI cloud
# client = OpenAI(api_key="sk-xxx")

# After: Ollama local (only 2 lines changed!)
client = OpenAI(base_url="http://localhost:11434/v1", api_key="ollama")

from openai import OpenAI

# Point to local Ollama server
client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama"  # Any non-empty string works
)

# Chat completion (identical to OpenAI API usage)
response = client.chat.completions.create(
    model="qwen2.5",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Explain RAG in 2 sentences"}
    ],
    temperature=0.3
)

print(response.choices[0].message.content)

# 执行成功

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama"
)

# Streaming works identically
stream = client.chat.completions.create(
    model="qwen2.5",
    messages=[{"role": "user", "content": "Write a short poem about AI"}],
    stream=True,
    temperature=0.7
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)
print()

# 执行成功

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama"
)

# Embeddings: just change the model name
response = client.embeddings.create(
    model="nomic-embed-text",  # Was: text-embedding-3-small
    input="What is the return policy for electronics?"
)

print(f"Embedding dimension: {len(response.data[0].embedding)}")
# 768 dimensions for nomic-embed-text

# 执行成功

from openai import OpenAI
import json

client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama"
)

# JSON mode: identical API
response = client.chat.completions.create(
    model="qwen2.5",
    messages=[
        {"role": "system", "content": "You are a product catalog API."},
        {"role": "user", "content": "List 3 laptops under $500"}
    ],
    response_format={"type": "json_object"},
    temperature=0.3
)

data = json.loads(response.choices[0].message.content)
print(json.dumps(data, indent=2))

# 执行成功

from openai import OpenAI
import json

client = OpenAI(
    base_url="http://localhost:11434/v1",
    api_key="ollama"
)

# Instead of Function Calling, use JSON mode + prompt
tools = {
    "get_order_status": {"order_id": "string"},
    "process_refund": {"order_id": "string", "amount": "number"},
    "search_products": {"query": "string", "category": "string"}
}

response = client.chat.completions.create(
    model="qwen2.5",
    messages=[
        {"role": "system", "content": f"""You are a customer service bot.
Available tools: {json.dumps(tools)}
If you need to call a tool, respond with JSON: {{"tool": "name", "args": {{...}}}}
Otherwise, respond normally."""},
        {"role": "user", "content": "Where is my order #12345?"}
    ],
    response_format={"type": "json_object"},
    temperature=0.2
)

result = json.loads(response.choices[0].message.content)
if "tool" in result:
    print(f"Call tool: {result['tool']} with args: {result['args']}")
    # Call: get_order_status with {"order_id": "12345"}
else:
    print("Direct response:", result)

Direct response:

# ============================================
# Comprehensive: SupportBot migration
# From GPT-4 to local Ollama with OpenAI SDK
# ============================================

from openai import OpenAI
import json
from dataclasses import dataclass
from typing import Optional

@dataclass
class SupportBotMigrator:
    """SupportBot with easy OpenAI/Ollama switching."""

    # Change these 2 lines to switch between cloud and local
    base_url: str = "http://localhost:11434/v1"
    api_key: str = "ollama"
    model: str = "qwen2.5"
    temperature: float = 0.4

    def __post_init__(self):
        self.client = OpenAI(
            base_url=self.base_url,
            api_key=self.api_key
        )
        self.system_prompt = (
            "You are SupportBot, an e-commerce customer service agent. "
            "Be polite, concise, and helpful. "
            "For order queries, ask for order number. "
            "If unsure, say 'Let me connect you with a human agent.'"
        )
        self.messages = [
            {"role": "system", "content": self.system_prompt}
        ]

    def chat(self, user_input: str) -> str:
        self.messages.append({"role": "user", "content": user_input})
        try:
            response = self.client.chat.completions.create(
                model=self.model,
                messages=self.messages[-10:],  # sliding window
                temperature=self.temperature
            )
            reply = response.choices[0].message.content
            self.messages.append({"role": "assistant", "content": reply})
            return reply
        except Exception as e:
            self.messages.pop()
            return f"Error: {e}"

    def classify_intent(self, user_input: str) -> dict:
        response = self.client.chat.completions.create(
            model=self.model,
            messages=[
                {"role": "system", "content": """Classify the customer intent.
Return JSON: {"intent": "order_status|refund|product_query|shipping|other", "confidence": 0.0-1.0}"""},
                {"role": "user", "content": user_input}
            ],
            response_format={"type": "json_object"},
            temperature=0.1
        )
        return json.loads(response.choices[0].message.content)

    def generate_embedding(self, text: str) -> list[float]:
        response = self.client.embeddings.create(
            model="nomic-embed-text",
            input=text
        )
        return response.data[0].embedding

# Usage - compare cloud vs local
if __name__ == "__main__":
    # Local Ollama (current config)
    bot = SupportBotMigrator()

    # Switch to OpenAI cloud (uncomment to use)
    # bot = SupportBotMigrator(
    #     base_url="https://api.openai.com/v1",
    #     api_key="sk-your-key",
    #     model="gpt-4"
    # )

    queries = [
        "Where is my order #88765?",
        "I want a refund for damaged goods",
        "Does this laptop have HDMI port?"
    ]

    for q in queries:
        intent = bot.classify_intent(q)
        reply = bot.chat(q)
        print(f"Q: {q}")
        print(f"Intent: {intent}")
        print(f"A: {reply[:100]}...")
        print()