How it works
Send the same prompt to multiple LLM instances simultaneously.
Parse each response into structured data using your schema.
Merge responses using majority voting and confidence scoring.
Basic usage
k-llms implements the K-LLMs methodology where multiple models evaluate the same prompt. Results are automatically reconciled into a single output with confidence scoring.
Works with any OpenAI-compatible models. Supports structured output validation with Pydantic models.
from k_llms import KLLMs
from pydantic import BaseModel
client = KLLMs()
class Event(BaseModel):
name: str
date: str
participants: list[str]
response = client.chat.completions.parse(
model="gpt-4.1-nano",
messages=[{"role": "user", "content": "Extract event info: Alice and Bob are going to a science fair on Friday."}],
response_format=Event,
n=4, # Run 4 parallel calls
)
print("Content:", response.choices[0].message.content) # Event dict as a json string
print("Parsed:", response.choices[0].message.parsed) # Event object
print("Likelihoods:", response.likelihoods) # Confidence scores
for i in range(1, len(response.choices)):
print(f"Choice {response.choices[i].index}: {response.choices[i].message.content}")
Why k-llms?
Enterprise-grade LLM consensus for everyone
k-llms gives you the state-of-the-art multi-model strategy popularized by Palantir's k-llm's approach with a single API call.
By using consensus voting across multiple models, you get a consistent result with higher accuracy, fewer hallucinations, and field-level confidence scores that help quantify uncertainty.
Watch: How k-llms unlocks reliability
A quick primer on the strategy that powers k-llms.