Point-by-point scores, prices and win-probability

A free tier plus three monthly plans: Basic $9.99, Pro $29.99 and Ultra $99.99. You can subscribe directly here for an instant API key, or on RapidAPI, Apify and API.market.

Yes — upgrade here. Upgrades apply immediately on your existing subscription and API key; Stripe bills only the prorated difference for the rest of the cycle. You never pay for two plans at once, and your key never changes.

Free: live scores & current matches, players, fixtures and your usage — 30 requests/min, 100/day; no history, odds, model fields or WebSocket. Basic ($9.99) adds historical match results, the point-by-point match tape (GET /history/matches/{id}), the results archive back to 1968, head-to-head (GET /h2h) and career aggregates — 60/min, 1,000/day. Pro ($29.99) adds match events, match-winner market odds and bulk monthly history packages — 300/min, 10,000/day. Ultra ($99.99) adds the AI win-probability model fields and analysis, per-match live statistics (aces, double faults, serve split, break points saved and converted), shot-level match-charting stats (serve placement, return depth, net & serve-and-volley, clutch break/game/set-point serving), head-to-head stat splits, the WebSocket live feed and webhooks — 600/min, 500,000/day.

Two layers, two depths. Results go back to 1968: the results archive holds 1,485,752 ATP and WTA results from 1968 through 2022 — final score, seeds, each player's rank at the time, round and draw metadata, and per-match serve statistics from 1991 — with yearly bulk packages. The point-by-point tape then takes over exactly where the archive ends: 44 monthly periods, January 2023 through August 2026, across ATP, WTA, Challenger and ITF — singles and doubles, 26,470,694 archived point-state rows over 171,323 of 174,969 completed matches, growing as the backfill runs. Bulk packages are published per calendar month (tape) and per year (archive), each with a sha256 manifest per file.

Yes, on ULTRA, and it is a different product from the tape. The tape gives the ordered score states of the match; rally construction gives how the point was played — serve direction, every stroke with its wing, direction and depth, the rally length and how the point ended. 11,638 charted matches (7,565 men’s back to 1960, 4,073 women’s), 1,850,490 points. Read one match at GET /rally/matches/{id} or take the whole thing as bulk yearly files. Charting is human work, so coverage is deep rather than universal and is concentrated on the biggest events — GET /rally/matches is the authoritative list of what exists.

Yes, on ULTRA, at three depths. Per live/completed match, GET /matches/{id}/statistics gives aces, double faults, the first/second-serve split, and break points saved and converted — and a completed match keeps its final figures, not just while it is live. Across a pairing, GET /h2h adds serve/return/break-point aggregates. And the deepest layer, GET /charting/players, gives shot-level career stats from the Match Charting Project — serve placement (deuce/ad × wide/body/T), return depth and outcomes, net and serve-and-volley conversion, and clutch break/game/set-point serving and returning — over 11,638 charted matches back to 1960. Coverage there is curated, majors-heavy, and every response reports its matches_charted sample.

The ordered score states of a completed match, with match metadata. Each match declares its own coverage, and the differences are real rather than cosmetic: from_start means we watched it live from 0-0 — every row carries a real timestamp, and our model's win-probability where it was computed (that stamp is best-effort, so where it was not, the field is null rather than guessed); partial means we picked it up mid-match; reconstructed means the point sequence came from a point-by-point feed rather than our own live observation, so those rows carry a null timestamp and no model fields. Nothing is ever synthesised from a final scoreline. The meta block on every response repeats that alongside raw_rows, unique_states and point_source, so you always know exactly what you are holding. Completeness is measured per match, not assumed: meta.points carries a measured verdict on every tape read, ?points=complete serves the sequence in full — score after each point, with server and point winner — where that verdict is complete, and ?points_complete=true filters the match list to measured-complete tapes. As of August 19, 2026, 75.0% of completed matches since 2023 are measured point-complete, and running backfills keep raising that share. Per-match tape reads come with Basic and above; bulk monthly packages come with Pro and above, or with a Historical Data API package purchase. There is no odds tape in the history product.

ATP, WTA, Challenger and ITF — both singles and doubles — from one arbitrated feed.

Yes. The Ultra plan streams score updates the instant a point is played, so you never have to poll. Continuous streaming runs on the push feed (GET /ws-token), built to fan out to thousands of subscribers; a native WebSocket remains available for quick console tests. Both carry per-point event frames on covered matches, with GET /matches/{id}/points for catch-up; each match declares its own pbp_coverage (point where a point vendor covers it, game otherwise), so you always know which cadence you are subscribed to.

Low latency. Scores come from an arbitrated multi-source feed and update within a second or two of the point being played.

Send your API key as an Authorization: Bearer header, or as X-API-Key. Keys are issued instantly when you subscribe.

A proprietary model's live win-probability for each player, with a confidence level and the key factors behind the call. It is available on the Ultra plan.

Yes — explore and test every endpoint in our public Postman workspace, and the /health endpoint needs no key at all.