Laya vs pplx-decider-flash

Laya

Apache-2.0 code and weights, installed with pip install laya, with Python 3.10 to 3.13 tested in CI. Base checkpoints score 0.362 and 0.352 on the maintainers' typed-decisions benchmark against a 0.318 random baseline, so it needs fine-tuning.

pplx-decider-flash

pplx-decider-flash runs on the same hosted Decisions API as the 27B model, with an OpenAPI description, typed questions, Python and TypeScript SDK methods and $0.01 per million input tokens, and its weights are public under Apache-2.0. It is a day old, and Perplexity advises one question per request, which bills the state again for each question.

Laya

1. Set LAYA_API_KEY before starting laya-serve. It listens on every interface by default
2. Gate on answer_confidence, not confidence, which measures entropy and doesn't match Jev's field
3. Shortlist choice questions with more than about 20 options using predict_shortlist or the laya_shortlist tool
4. Use semantic or opaque labels such as A and B, not yes and no, in choice questions. The checkpoints can follow the label text
5. Pass model="multilingual" and max_len=8192 for long documents. The English checkpoint stops at 512 tokens

pplx-decider-flash

1. Send Authorization: Bearer with model set to pplx-decider-flash-v1.2. A key in x-api-key returns 401.
2. Send one question per request by default. Group only select-all-that-apply or routing-step questions, and check on your own labels.
3. Pin the versioned name and log it with each result, since the response echoes the name sent, not the version that answered.
4. Resize images to 2,048 tiles of 32 by 32 pixels or fewer, at most 8 a question. Larger images end in a 504 after about a minute.
5. Compare it with pplx-decider-v1.2 on a labelled sample at your thresholds before choosing it.