Skip to main content

​
Usage

from langchat.providers import OpenAI

llm = OpenAI("gpt-4o-mini")              # reads OPENAI_API_KEY
llm = OpenAI("gpt-4o", temperature=0.3)  # explicit model + temp
Pass to LangChat: 
from langchat import LangChat
from langchat.providers import OpenAI, Pinecone, Supabase

lc = LangChat(
    llm=OpenAI("gpt-4o-mini"),
    vector_db=Pinecone("my-index"),
    db=Supabase(),
)

​
Parameters

​
model
str
default:"gpt-4o-mini"
OpenAI model name. First positional argument.
​
api_key
str | None
default:"None"
Single API key. Falls back to OPENAI_API_KEY environment variable.
​
api_keys
list[str] | None
default:"None"
Multiple API keys for automatic rotation. Takes precedence over api_key.
​
temperature
float
default:"1.0"
Sampling temperature. 0.0 = deterministic. 1.0 = creative.
​
max_retries_per_key
int
default:"2"
Number of retries per key before rotating to the next one.

​
API key rotation

When you have multiple API keys (e.g., to handle rate limits across multiple OpenAI projects), pass them as a list: 
llm = OpenAI(
    "gpt-4o-mini",
    api_keys=[
        "sk-proj-key1...",
        "sk-proj-key2...",
        "sk-proj-key3...",
    ],
    max_retries_per_key=2,
)
If a request fails on key1, LangChat retries up to max_retries_per_key times, then rotates to key2, and so on. Total maximum attempts = len(api_keys) × max_retries_per_key.

​
Environment variable

Set OPENAI_API_KEY in your .env: 
OPENAI_API_KEY=sk-...

​
Available models

ModelContextSpeedQuality
gpt-4o-mini128kFastGood — recommended default
gpt-4o128kMediumHigh
gpt-4-turbo128kMediumHigh
gpt-3.5-turbo16kFastestBasic
Any valid OpenAI chat completion model name is accepted.