Parallel Decisions API searches the live web for typed answers at $1 per 1,000 calls. See how citations, refusals and its benchmark work.
By Parallel · Released 25 Oct 2026
A decision engine that searches the live web before it answers. You send typed questions about a company, product or policy, and one call returns typed answers with probabilities (or a refusal) plus a shared list of the web pages it used, so it can answer things a model only knows from training data cannot, such as whether a company has an open job right now.
Parallel Decisions API searches the live web and answers typed questions about the shared state. That suits facts that can change, such as whether a company offers a product or has an open job. The response can contain a refusal, which calling code must handle. Citations belong to the whole request, rather than to individual answers.
Parallel lists $1 per 1,000 calls and 5,000 free requests to start. Several questions can share one state in one paid call. This is a per-call price; hosted models priced by input tokens use a different billing unit.
Parallel reports 76.22% on its own 300-question PWS-Decisions benchmark. It used live search, while the compared decision models had no web access or supplied external context. The benchmark tests current web facts; it does not establish a lead on questions where every model receives the same evidence. Dataset terms prohibit redistribution.
Parallel launched Decisions Fast on OpenRouter on 6 Nov 2026 with a seven-day free promotion. On 7 Nov the decisions inventory lists parallel/decisions-fast:free with 32,000 tokens and text input; the unsuffixed route present in the morning snapshot is absent. An inventory omission does not prove that requests to that route fail. The promotional zero price is separate from the direct API's recorded per-call rate. Parallel says up to 100 related questions can share one search call.
Searches the live web per request; citations cover the whole call, not each answer
Latency
VENDOR-REPORTED: 0.56 seconds median on PWS-Decisions, measured by Parallel on 29 Oct 2026 (Pacific).
Claims and sources
Parallel says it costs 10x less than a standalone model vendor reportedsource
Parallel reports 76.22% on its own 300-question PWS-Decisions set with live web search, against Perplexity Decider v1.2 at 50.89%, OpenAI's Decisions API at 41.67% and Jev 1.14 at 33.44%, run without web access or supplied external context. Uniform random guessing is 42.42%. LLMs with native search scored 96.56% to 97.33% at about 16.5 to 28.8 times the cost. Parallel generated these questions from its own customer topics; dataset terms prohibit redistribution. vendor reportedsource
The benchmark blog chart labels Baseten Elephant at 32.4% overall and 32.3% public. The repository records Baseten Mammoth at 45.0% overall and 37.0% public. vendor reportedsource
A small demo with a live site: type a US or Indian stock and two Parallel Decisions calls, through Parallel's free MCP endpoint, assess its technicals and trading activity from the live web and return probabilities and sources.