# Which search? (/docs/choosing-endpoints/search) One platform, the forums, everywhere, several at once, or a stored topic: how to pick. Costs are per successful call. Failed and cached responses are not charged. ## Search one platform, the forums, or everywhere? [#search] To search one platform, call that platform's own search. To search several platforms over a date window with every row judged and de-duplicated, call `/v1/prism/discover`. For a quick fused sweep at a flat 20 credits, call `/v1/search/everywhere`, which drops rows outside `from_date`, `to_date`, `since` or `until` but reads one page per source, so a past window can come back thin. The table below has the per-platform searches and the specialised cross-platform ones. | Your goal | Use this endpoint | Why | Credit cost | | ---------------------------------------------------------- | -------------------------------------------------------------- | ----------------------------------------------------------------------------------------- | ----------------------------------------- | | Search a platform you already chose | `/v1/tiktok/search`, `/v1/youtube/search`, `/v1/reddit/search` | Single-platform keyword search | 1cr (YouTube 1-11cr with `include`) | | Find creators by keyword, with bio and region on every row | `/v1/tiktok/search/users` with `include=profile` | One call instead of user search plus a profile lookup per row; unfilled rows refunded | 1cr, +1 per row filled (31 at most) | | Search inside one subreddit | `/v1/reddit/subreddit/search` | Scoped to a named community | 1cr | | Find which communities discuss a topic, with their size and activity | `/v1/reddit/subreddits/search` with `include=details` | One call instead of community search plus a details lookup per community; unfilled rows refunded | 1cr, +1 per community filled (26 at most) | | One keyword across all of Reddit, with comments | `/v1/reddit/omni-search` | Runs the search, expands the top threads' comments, rolls up which subreddits are talking | 5cr min, up to 9cr (metered by `threads`) | | Fused forum search (Reddit + Hacker News + Naver) | `/v1/search/forums` | One ranked list across forum sources, each thread with a stance and a judged relevance score | 10cr | | Multi-country news in one call | `/v1/search/news` | LLM-planned angles, localized per Google News edition, deduped across legs, grouped into same-event stories | 2-14cr (metered) | | One query across 18 social platforms | `/v1/search/everywhere` | LLM-planned, RRF-fused, reranked, clustered, each result with a stance and a relevance score | 20cr flat | | Several platforms over a date window, every row judged | `/v1/prism/discover` | Keeps only rows from your dates, judges every row against your question, removes duplicates, and continues in the background when the request is large ([guide](/docs/prism/discover.md)) | Each platform's own page price, plus any judging against a `context` or a `label` | Start with per-platform `search` (1cr) when the platform is decided. Fourteen separate 1cr calls still leave the merge and the date window to you, which is what `/v1/prism/discover` does for each platform's own page price, plus any judging against a `context` or a `label`. The post searches (`/v1/tiktok/search`, `/v1/youtube/search`, `/v1/reddit/search` and their siblings on other platforms), `/v1/search/forums` and `/v1/search/everywhere` score each row's relevance to your query by default, free, on `computed.relevance`. When your keyword is also a common word or another brand's name, add `relevance=filter` to drop the rows that are about something else, still at no extra credit. When you want to know how people feel rather than what they posted, `/v1/search/everywhere` and `/v1/search/forums` count stances for you (`data.stance_split`). See [Labels](/docs/labels.md). **The four Reddit search endpoints are a ladder, not alternatives** `/v1/reddit/subreddits/search` (1cr) is the step before the other three: it answers which communities discuss a subject, and `author.id` on each row is the `subreddit` parameter the rest of the platform takes. Send `include=details` and each community also arrives with its creation date, weekly active users, weekly contributions, rules and language, at 1 credit per community filled and 26 at most, which is the same price as the per-community lookups it replaces. `/v1/reddit/search` (1cr) is a plain site-wide post list. `/v1/reddit/subreddit/search` (1cr) narrows the same job to one community. `/v1/reddit/omni-search` (5cr floor) runs that search _and_ expands the top threads' comment trees in parallel _and_ returns per-subreddit volume with a tone label, so it is never the cheap option. Take it when you want voice-of-customer output rather than a list of links. The `threads` parameter is what moves it between 5 and 9 credits. **Source names on /v1/search/everywhere are not platform names** Unknown source names are dropped silently at full price. The X source is `twitter-ai-search`, not `twitter`, so `sources=reddit,twitter,tiktok` succeeds, quietly drops X coverage, and still charges the flat 20 credits. Check `data.sources_called` in the response to confirm the fan-out you asked for, and see the [full source table](/docs/search/everywhere.md#sources) for every canonical name. ## Which call runs one fused search? [#fused-search] For one fused, ranked search across the platforms it covers at a flat 20 credits a search, call `/v1/search/everywhere`; for a collection of more than 100 rows or more than one page per platform, call `/v1/prism/discover`; for creators, `/v1/prism/creators`; to be told what is new, send `since_watermark` with the `data.next.watermark` of your last answer, or turn the search into a watch with its `data.next.watch`. See [Universal search](/docs/search/everywhere.md) for the agent path. ## When a search should become a watch [#search-to-watch] If you would run the same search again tomorrow, do not loop it. Turn it into a watch so you receive only what is new. [Which call tells me when something is new?](/docs/choosing-endpoints/watching.md) picks the watch, and [Webhooks and events](/docs/webhooks-and-events.md) shows how the result reaches you. ## How much does one page of search cost? [#search-cost] Every search endpoint's credit cost, including the range of each metered one, is in [Endpoint pricing](/docs/endpoint-pricing.md). The table above prints the cost beside each call, and the pricing page is the source if the two ever differ.