Parallel Decisions now available on AI Gateway for free
Parallel Decisions Fast adds web-grounded Boolean, Choice, and Score answers to AI Gateway, with probabilities and citations. Its citations cover the request as a whole, so builders should not treat them as source attribution per answer.
Parallel Decisions Fast searches current web information and returns probabilities with cited evidence. It supports **Boolean, Choice, and Score** questions, and related questions can share context and retrieved evidence in one request. It is free through **November 23, 2026** on AI Gateway. Builders can use it for tasks such as checking eligibility or verifying claims, then route uncertain results for review. The model is available as `parallel/parallel-decisions-fast`; no separate Parallel account is required. Citations apply to the whole request rather than mapping sources to individual answers. The material describes automatic action above a probability threshold as an option, but does not specify a generally safe threshold.
No separate implication has been recorded yet.
Parallel Decisions Fast searches current web information and returns probabilities with cited evidence. It supports **Boolean, Choice, and Score** questions, and related questions can share context and retrieved evidence in one request. It is free through **November 23, 2026** on AI Gateway. Builders can use it for tasks such as checking eligibility or verifying claims, then route uncertain results for review. The model is available as `parallel/parallel-decisions-fast`; no separate Parallel account is required. Citations apply to the whole request rather than mapping sources to individual answers. The material describes automatic action above a probability threshold as an option, but does not specify a generally safe threshold.
Parallel Decisions adds a structured way to turn current web evidence into Boolean, Choice, or Score results with probabilities, rather than returning search material alone. Its request-level citations limit answer-level traceability, and the free window makes experimentation easier without resolving what probability should trigger action.