meraGPT
meraGPT

Decider 1

Release Date Unknown4K INProprietary Model

Decider 1 is meraGPT's decision model for returning probability distributions over yes/no questions, labels and scoring rubrics from supplied text state. It answers multiple questions together through the System One schema, but cannot write open-ended responses or retrieve facts absent from the input. AIB places classification in the weaker band at the time checked; the vendor's comparison with Jev uses teacher-model answers as its reference, not universal ground truth. Short combined limits for state and questions and a cap on choice options require splitting long documents and organizing large label sets hierarchically. The weights are proprietary and output is unbilled, while the model alias can move to an improved version, so validate thresholds on the actual languages and workflows and recheck after updates.

API cost

$0.04IN$0OUT

Expert Performance Evaluation

54.747.7

More about AIB DecisionBench-KJ

See each evaluation’s detail page for its source and last updateData corrections & removal requests