Tsuzuki API v1 · FREE · NO KEY

A read-only REST API for anime airing schedules. Unlike the upstream data it is built on, every episode comes back with the release variants that actually exist — the Japanese broadcast, the subtitled simulcast and the English dub — and with human corrections for delays and broadcast breaks already applied.

Why not just call AniList?

You should, for most things — this API is built on AniList and credits it everywhere. But AniList publishes one time per episode: the Japanese TV broadcast. It has no field for when Crunchyroll actually posts the sub, no dub dates at all, and no way to say "this week's episode was pre-empted by a sports block."

That correction layer is what this API adds. If you only need Japanese broadcast times, go straight to AniList and skip the hop.

Conventions

The air-type model

Every episode entry carries an airType and two honesty flags:

A dub entry only exists where someone has confirmed a dub date. We would rather return nothing than a date we invented — so an absent dub means "unknown", not "no dub".

Endpoints

GET /api/v1

Service description and the machine-readable endpoint list.

GET /api/v1/schedule

Corrected schedule for a date window. This is the endpoint most people want.

curl "https://tsuzuki.top/api/v1/schedule?days=1&airType=sub&platform=Crunchyroll"

{
  "ok": true,
  "query": { "start": "2026-08-04", "days": 1, "airType": "sub", "platform": "Crunchyroll" },
  "count": 12,
  "episodes": [
    {
      "mediaId": 128757,
      "episode": 5,
      "airType": "sub",
      "airingAt": 1754321400,
      "airingAtIso": "2026-08-04T13:30:00.000Z",
      "exact": true,
      "estimated": false,
      "platform": "Crunchyroll",
      "isBreak": false,
      "note": null,
      "title": "Young Ladies Don't Play Fighting Games",
      "coverImage": "https://s4.anilist.co/…"
    }
  ],
  "attribution": "Data from AniList, corrected by Tsuzuki."
}

GET /api/v1/anime/{anilistId}

One title with its full per-episode variant schedule, plus the raw corrections record if one exists. Break weeks are included here. Add ?full=1 for the untouched media record instead of the summary shape.

curl "https://tsuzuki.top/api/v1/anime/128757"

GET /api/v1/seasons/{season}/{year}

A season's lineup, each title with its nextEpisode already resolved through the correction layer. season is winter, spring, summer or fall. ?full=1 returns the complete media records — that is what the Tsuzuki app itself loads.

curl "https://tsuzuki.top/api/v1/seasons/summer/2026"

GET /api/v1/airing

Currently-airing titles that AniList never assigned a season to — Korean and Chinese productions, and long-running ONAs that don't map onto a Japanese broadcast season. A season query cannot reach these by construction, so if you build a calendar the way everyone does (three seasons calls around the date you care about) you will silently miss them. They are already folded into /schedule; this endpoint exists for consumers that page season by season. ?full=1 behaves as it does on seasons.

curl "https://tsuzuki.top/api/v1/airing"

GET /api/v1/franchise/{anilistId}

Every anime in the same franchise, in release order. Upstream only ever gives a title its direct relations, so asking season 3 what it belongs to gets you seasons 2 and 4 and nothing else. This walks the whole relation graph and returns the lot — seasons, movies, OVAs, specials, spin-offs — each with the relationLabel that placed it there and isRoot on the title you asked about.

Entries are sorted by start date. One known only to the year sorts after everything precisely dated in that year, and one with no date at all sorts last. Very large franchises are capped; truncated tells you when that happened.

curl "https://tsuzuki.top/api/v1/franchise/16498"

GET /api/v1/search

Title search. q is a fragment of any title (English, romaji or native), limit is 1–25 and defaults to 12. Answers from the seasons already cached where it can and reaches upstream only for titles outside them — source in the response says which happened. ?full=1 returns complete media records.

curl "https://tsuzuki.top/api/v1/search?q=frieren"

GET /api/v1/overrides

The raw correction document — every offset rule, exact time and delay/break note we hold, keyed by AniList id. Useful if you want to apply the corrections yourself against data you already have. The schema is documented inside /data/overrides.json.

Errors