Skip to main content

​
Installation

pip install meshai-sdk[openai]

​
Usage

from meshai import MeshAI
from meshai.integrations.openai import wrap_openai
import openai

# Initialize MeshAI
client = MeshAI(api_key="msh_...", agent_name="my-agent")
client.register(framework="custom", model_provider="openai")

# Wrap the OpenAI client
oai = wrap_openai(openai.OpenAI(), meshai=client)

# Use as normal — all calls auto-tracked
response = oai.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello"}],
)

# Model, tokens, and cost are captured automatically
# Even when switching models:
response = oai.chat.completions.create(
    model="gpt-4o-mini",  # Different model — tracked separately
    messages=[{"role": "user", "content": "Hello"}],
)

​
What’s Tracked

For each API call, MeshAI captures:
  • Model name (extracted from the response, not hardcoded)
  • Input tokens (prompt_tokens)
  • Output tokens (completion_tokens)
  • Request type (chat.completions)

​
Alternative: Proxy (Zero-Code)

If you prefer no SDK at all: 
export OPENAI_BASE_URL=https://proxy.meshai.dev/v1/openai/k/msh_YOUR_PROXY_KEY
All OpenAI calls are automatically routed through MeshAI.