Skip to content

GQI

GQI predicts a number from text, code, logs, and mixed structured input.

Provide labeled (X, y) examples, fit private task-specific state, and use the immutable returned model handle for new predictions. Every finite target is accepted, including zero and negative values.

Python quickstart

Create a key in the account console, then install the dependency-free client:

pip install gqi-labs

from gqi import GQI

client = GQI(api_key="gqi_sk_...")

X = [
    "checkout fails for every user", "minor copy edit", "refund is stuck",
    "password reset email delayed", "production API is down", "feature request",
    "invoice has the wrong address", "documentation link is broken",
]
y = [10, 1, 7, 4, 10, 2, 5, 2]

fitted = client.fit(task_name="ticket-priority", X=X, y=y)
result = client.predict(
    model=fitted.model,
    X=["payments fail for every customer"],
    config={"num_samples": 16, "return_samples": True},
)
print(result.rows[0].prediction, result.rows[0].iqr)

What the API provides

  1. Explicit model choice. Prediction requires either the immutable fitted handle or the literal "GQI" for an intentional zero-shot call. Omitting the model never silently becomes an untuned prediction.
  2. Sample-based output. Prediction returns the median, interquartile range, and optionally every finite sample.
  3. Managed fitting. The completed fit reports its actual row split, steps, context length, loss, and truncation count.
  4. Asynchronous jobs. Fit and predict can be polled, retried idempotently, and cancelled.

Where to go next