RTP VERSION TRACKER

Not sure which to choose? See what we recommend for each type of player.

Which Sweet Bonanza RTP does each casino run?

Every engine we could read ran the same configuration: 96.49% for the base game and 96.60% in Ante Bet mode, at Gamdom, Rainbet, Stake and 7Bit on 17 September 2026. Each launch carried its own operator style name (for example gmdm_gamdom for Gamdom and rare_stake for Stake), so the four readings came from four separate casino launches, not one page loaded four times. BitStarz is the one casino whose figure differs: its paytable said 96.52% two days earlier, and it could not be re-checked at engine level because the game page then asked for a login.

How is each reading taken?

A reading starts on the casino’s own game page, in demo or fun mode, or by opening the demo frame’s address directly. For Pragmatic Play titles the game client then sends an initialisation request to the provider’s game server (a path ending in /gs2c/ge/v3/gameService; the initialisation call is doInit). Its response carries the session settings, and one of them is rtp=, the return the maths engine has been told to run for that launch. For Sweet Bonanza it reads rtp=96.49,96.60: base game first, Ante Bet second. We read it from the browser’s network log and record the casino, the game, the value and the date.

Where the engine cannot be seen, the next source is the game’s rules or paytable screen, and below that the casino’s lobby label. Every figure carries one of three tiers:

• Tier A, the game’s own configuration. The engine init value or the in-game paytable line. Authoritative for the build that launched.
• Tier B, the casino’s page. A figure printed in the lobby or on the game page. It is what the casino claims and can lag the build.
• Tier C, the provider’s figure. The number on the studio’s own game page. A reference line, never a reading for any casino.

The step-by-step version, for checking a game yourself, is in how to check a slot’s RTP at your casino.

What we did not do

• No real-money bets, deposits or accounts. Every engine reading is a logged-out demo launch, which is why login-gated casinos have empty rows.
• No RTP was estimated from spin results. A few hundred demo spins say nothing reliable about a figure measured in hundredths of a point.
• No figure was copied from another site, an aggregator database or a slot review. The provider “default” column of the tracker’s earlier version on HighRTPHub was not carried over, because those figures were not readings.
• No attempt to decrypt aggregator traffic. Where a casino launches through an encrypted aggregator, the row says so.
• One slot so far. Sweet Bonanza and its 1000 edition are the only games with readings.

What do the numbers mean?

Label against engine. Stake’s 96.50% against an engine figure of 96.49% is a difference of one hundredth of a point: $1 of expected loss per $10,000 wagered ($351 at 96.49%, $350 at 96.50%). Nobody loses money over that. The point is what the gap shows: the lobby label is written by people and can drift from the game, while the init value comes from the system that settles the spins. When the two disagree, only one of them is the configuration.

Ante Bet. The 96.60% Ante Bet figure is higher than the 96.49% base, but the mode is not cheaper per spin. The BitStarz paytable (15 September 2026) listed the bet multiplier as 20x normally and 25x with Ante Bet. Expected cost per spin is the stake times the house edge: 20 × 3.51% = 0.70 coins without Ante Bet and 25 × 3.40% = 0.85 coins with it, about 21% more per spin for a better return per coin staked.

Identical engines are weaker evidence than different ones. Gamdom, Rainbet and Stake were all served from the same demo host, and their responses carried the same jurisdiction=99 setting. A demo host may hand every operator the same default, so four matching values do not prove four matching real-money builds. A casino whose value differs, as BitStarz’s paytable did, is the more informative case, and it is the first row to re-check.

Provider page against engine. Pragmatic Play’s own Sweet Bonanza page lists 96.48% (read 29 October 2026), one hundredth below every engine we read. We have not found out why, and we do not treat either number as a correction of the other. For Sweet Bonanza 1000, Duel’s page label implies 96.55%, two hundredths above the 96.53% on Pragmatic Play’s page; the game itself was not read.

Does the same slot pay a different RTP at different casinos?

It can, and the readings so far have not caught a confirmed case. Four engines agreed exactly; one paytable (BitStarz, 96.52%) disagreed and is waiting for a re-check. Pragmatic Play’s two game pages show a single RTP each and say nothing about other versions (read 29 October 2026), so the provider page cannot tell you which build a casino runs. What can is the game itself. In Great Britain the Gambling Commission’s Remote Technical Standard 3C lists the RTP percentage among the information that must be easily available before a customer commits to gamble (page last updated 7 May 2024, read 29 October 2026). Offshore crypto casinos are not bound by that standard, which is why the engine reading matters most there.

Readings were made from Finland. Operators can serve different builds by region, so a player elsewhere may see a different figure.

Gaps: casinos we could not read

Change log