Overview / Description
Decisions API is an AI developer tools platform that classifies, scores, and routes text through a single API with structured, probabilistic outputs. Instead of a generative model, it exposes 11 specialized decision models (named Jev, Liquid d1, Kev, Solar, Tev1, Span, and others) that return typed answers to three question kinds: choice, score, and noul (yes/no). Every answer carries a probability estimate, so downstream code can act on confidence rather than a raw string. The homepage positions support-ticket triage and intelligent routing as the flagship use case — reading an incoming message and directing it to the right team with a confidence score — and lists content moderation and lead scoring as adjacent fits. Developers can test prompts in a browser-based playground before wiring the API in, with a single authentication model spanning both web and API, plus request tracking and credit usage visibility. For teams that need interpretable, structured decisions rather than free-form generation, that typed-output design is the point. On BestAIFor we list Decisions API under developer tools because the integration surface is an API and a playground for engineers, not an end-user app.
Used For
Classifying, scoring, and routing text in support triage, content moderation, and lead scoring workflows.
Pricing
Plan
$1 = 10,000 credits on every pack; model rates 300-720 credits per million input tokens, minimum 1 credit per request
Pros & Cons
Pros
- Three typed question kinds — choice, score, and noul (yes/no) — with a probability estimate on every answer
- 11 specialized decision models accessible through one unified API
- Browser-based playground to test prompts before integrating the API
- Single authentication across web and API, with request tracking and credit usage visibility
- Credits do not expire, and packs scale from $10 to $1,000
Cons
- Built for classification, scoring, and routing — not generative or conversational use
- Requires developer integration; no no-code or end-user interface beyond the playground
- Model names (Jev, Kev, Solar, etc.) are opaque without reading the per-model credit rates
- Credit math (300-720 credits per million input tokens) takes some work to estimate cost
Questions & Answers
Alternatives
OpenAI moderation and classification endpoints, Cohere Classify, and Google Cloud Natural Language.
Reviews & Ratings
0 reviews
Sign in to rate and review Decisions API.
Sign in to reviewNo reviews yet. Be the first to review Decisions API!