Ravelin benchmark
Ravelin, the daily tower defense, played with the game's own orders while the clock is paused. The bench runs the game's own simulation, so the boards are the ones people play: the daily sieges of fixed dates, campaign levels, and Iron levels the game's own bot cannot hold. A battle scores against par, the game's reference clear on the same board. Compare wins, lives lost, refused orders, and the cost and time of a decision.
Ravelin leaderboard
Select a column to sort. Scroll sideways for all metrics →
| # | Model | Score | 95% interval | Score per tierDayCmpIrn | Won | Beat par | Lives lost | Refused | Decisions | Battles | API cost / battle | Sec / decision |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Gemini 4 Neonreasoning medium | 107 | 13313157 | 5 of 6 | 5 of 6 | 0.2 | 0.6% | 11 | 6 | $0.887 | 40.4 | |
| 2 | Gemini 4 Hydrogenreasoning medium | 99 | 12011957 | 5 of 6 | 5 of 6 | 1.7 | 0.7% | 11 | 6 | $0.327 | 16.6 | |
| 3 | GPT-6.2 Astrareasoning medium | 89 | 13012413 | 4 of 6 | 4 of 6 | 1.3 | 4.7% | 10 | 6 | $0.323 | 13.6 | |
| 4 | GPT-6.1 Solreasoning medium | 87 | 1391167 | 4 of 6 | 4 of 6 | 0.5 | 4.3% | 9 | 6 | $0.138 | 14.0 | |
| 5 | GPT-6.1 Terrareasoning medium | 86 | 13610913 | 4 of 6 | 4 of 6 | 1.0 | 1.4% | 10 | 6 | $0.047 | 7.2 | |
| 6 | GPT-6 Solreasoning medium | 83 | 1351085 | 4 of 6 | 4 of 6 | 0.8 | 0.7% | 9 | 6 | $0.140 | 16.0 | |
| 7 | GPT-6 Astrareasoning medium | 82 | 1301125 | 4 of 6 | 4 of 6 | 0.8 | 7.8% | 9 | 6 | $0.734 | 12.2 | |
| 8 | Qwen3.8 Flash-Nextlocal, thinking low · open weights | 73 | 131853 | 4 of 6 | 3 of 6 | 4.3 | 0.8% | 11 | 6 | n/a | 587.5 | |
| 9 | Claude Haiku 5.5reasoning medium | 72 | 1111023 | 4 of 6 | 3 of 6 | 2.2 | 0.0% | 10 | 6 | $0.013 | 9.9 | |
| 10 | Claude Sonnet 5.5reasoning medium | 70 | 109975 | 4 of 6 | 2 of 6 | 3.8 | 0.0% | 12 | 6 | $0.122 | 5.0 | |
| 11 | GPT-6 Lunareasoning medium | 57 | 73953 | 3 of 6 | 2 of 6 | 6.0 | 15.4% | 11 | 6 | $0.0080 | 9.8 | |
| 12 | Gemini 4 Argonreasoning medium | 843 of 6 | 1231256 | 2 of 3 | 2 of 3 | 3.0 | 1.4% | 10 | 3 | $0.129 | 9.7 | |
| 13 | Qwen3.8 27Blocal, no thinking · open weights | 383 of 6 | 17926 | 1 of 3 | 0 of 3 | 10.0 | 37.0% | 12 | 3 | n/a | 148.3 | |
| 14 | Ministral 3 8Blocal, no thinking · open weights | 253 of 6 | 38316 | 0 of 3 | 0 of 3 | 13.7 | 62.1% | 16 | 3 | n/a | 40.7 | |
| 15 | Qwen3.5 9Blocal, no thinking · open weights | 173 of 6 | 21256 | 0 of 3 | 0 of 3 | 13.7 | 43.1% | 12 | 3 | n/a | 34.9 | |
| 16 | Gemma 3 4Blocal, no thinking · open weights | 143 of 6 | 17196 | 0 of 3 | 0 of 3 | 13.7 | 88.1% | 12 | 3 | n/a | 37.9 |
No models match your search.
Score is 100 times the mean points over every battle a model played: a win scores par divided by its score (battle time plus 5 seconds per life lost), never below 0.5; a loss scores half the share of the waves it held. A model ranks once it has played the first 2 battles of every tier; until then it shows how many of those it has played and ranks after the models that have. Past that the bench keeps playing further battles in the same order for every model, and every finished one counts, so n grows and the intervals narrow.
The Pareto frontier shows where no other model scores at least as well for less cost or time, with one strict improvement. Partial sets are included and may cover different items. Use completed sets for direct comparisons.
By tier
Score against par with its 95% interval, then battles won, lives lost per battle, and how a won battle's score compared with par (below 1 is faster). The count after each result is n, the battles finished in that tier.
| Model | Daily siege | Campaign | Iron |
|---|---|---|---|
| Gemini 4 Neonreasoning medium | 133 (99 to 166)n 2 · 2 won · 0.0 lives lost · 0.77x par when won | 131 (100 to 163)n 2 · 2 won · 0.0 lives lost · 0.77x par when won | 57 (0 to 156)n 2 · 1 won · 0.5 lives lost · 0.93x par when won |
| Gemini 4 Hydrogenreasoning medium | 120 (110 to 130)n 2 · 2 won · 4.0 lives lost · 0.83x par when won | 119 (116 to 121)n 2 · 2 won · 0.5 lives lost · 0.84x par when won | 57 (0 to 159)n 2 · 1 won · 0.5 lives lost · 0.92x par when won |
| GPT-6.2 Astrareasoning medium | 130 (103 to 158)n 2 · 2 won · 3.0 lives lost · 0.78x par when won | 124 (118 to 130)n 2 · 2 won · 0.0 lives lost · 0.81x par when won | 13 (0 to 31)n 2 · 0 won · 1.0 lives lost |
| GPT-6.1 Solreasoning medium | 139 (111 to 167)n 2 · 2 won · 0.0 lives lost · 0.73x par when won | 116 (105 to 128)n 2 · 2 won · 0.5 lives lost · 0.86x par when won | 7 (4 to 10)n 2 · 0 won · 1.0 lives lost |
| GPT-6.1 Terrareasoning medium | 136 (118 to 154)n 2 · 2 won · 2.0 lives lost · 0.74x par when won | 109 (95 to 123)n 2 · 2 won · 0.0 lives lost · 0.92x par when won | 13 (0 to 31)n 2 · 0 won · 1.0 lives lost |
| GPT-6 Solreasoning medium | 135 (94 to 176)n 2 · 2 won · 0.0 lives lost · 0.76x par when won | 108 (101 to 116)n 2 · 2 won · 1.5 lives lost · 0.92x par when won | 5 (4 to 6)n 2 · 0 won · 1.0 lives lost |
| GPT-6 Astrareasoning medium | 130 (95 to 165)n 2 · 2 won · 1.0 lives lost · 0.79x par when won | 112 (108 to 115)n 2 · 2 won · 0.5 lives lost · 0.90x par when won | 5 (4 to 6)n 2 · 0 won · 1.0 lives lost |
| Qwen3.8 Flash-Nextlocal, thinking low | 131 (96 to 166)n 2 · 2 won · 1.0 lives lost · 0.78x par when won | 85 (55 to 115)n 2 · 2 won · 11.0 lives lost · 1.21x par when won | 3 (0 to 8)n 2 · 0 won · 1.0 lives lost |
| Claude Haiku 5.5reasoning medium | 111 (92 to 130)n 2 · 2 won · 4.5 lives lost · 0.91x par when won | 102 (86 to 117)n 2 · 2 won · 1.0 lives lost · 0.99x par when won | 3 (0 to 8)n 2 · 0 won · 1.0 lives lost |
| Claude Sonnet 5.5reasoning medium | 109 (93 to 125)n 2 · 2 won · 5.5 lives lost · 0.92x par when won | 97 (92 to 102)n 2 · 2 won · 5.0 lives lost · 1.03x par when won | 5 (4 to 6)n 2 · 0 won · 1.0 lives lost |
| GPT-6 Lunareasoning medium | 73 (0 to 175)n 2 · 1 won · 11.0 lives lost · 0.80x par when won | 95 (82 to 108)n 2 · 2 won · 6.0 lives lost · 1.06x par when won | 3 (0 to 8)n 2 · 0 won · 1.0 lives lost |
| Gemini 4 Argonreasoning medium | 123n 1 · 1 won · 8.0 lives lost · 0.82x par when won | 125n 1 · 1 won · 0.0 lives lost · 0.80x par when won | 6n 1 · 0 won · 1.0 lives lost |
| Qwen3.8 27Blocal, no thinking | 17n 1 · 0 won · 20.0 lives lost | 92n 1 · 1 won · 9.0 lives lost · 1.08x par when won | 6n 1 · 0 won · 1.0 lives lost |
| Ministral 3 8Blocal, no thinking | 38n 1 · 0 won · 20.0 lives lost | 31n 1 · 0 won · 20.0 lives lost | 6n 1 · 0 won · 1.0 lives lost |
| Qwen3.5 9Blocal, no thinking | 21n 1 · 0 won · 20.0 lives lost | 25n 1 · 0 won · 20.0 lives lost | 6n 1 · 0 won · 1.0 lives lost |
| Gemma 3 4Blocal, no thinking | 17n 1 · 0 won · 20.0 lives lost | 19n 1 · 0 won · 20.0 lives lost | 6n 1 · 0 won · 1.0 lives lost |
The game's bot on the same boards
| Player | Score | Daily siege | Campaign | Iron |
|---|---|---|---|---|
| Par (reference bot)The game's own reference bot at its own pace. Its score on a board is the par a player sees. | 764 of 6 won | 1002 won | 1002 won | 280 won |
| Rusher botThe same bot calling every wave the moment it may, for the fastest clears. | 1144 of 6 won | 1482 won | 1652 won | 300 won |
| Bot in paused timeThe same bot held to the bench's decision points, as many orders as it likes at each. | 583 of 6 won | 641 won | 902 won | 190 won |
| Novice botThe same bot playing plainly: no abilities, never calls a wave early. | 634 of 6 won | 972 won | 852 won | 70 won |
Every board
Every battle
One square per board, filled when the battle held every wave. Columns run tier by tier, in the order every model meets them.
MethodologyProtocol, scoring & limitations
The boards
Ravelin is a tower defense on a war table: attackers march along the road from the gates, and towers built beside it shoot them before they reach the keep. Each attacker that gets through costs lives. The bench runs Ravelin's own simulation, pinned to commit 23c8ab1, so a later balance change in the game cannot move anyone's result. Every battle stores the game's own tick tape, and the same board and orders reproduce the same battle, in the bench's driver and in the game's own replayer.
| Tier | Boards | Items | Reference bot | Rusher bot |
|---|---|---|---|---|
| Daily siege | The daily board of a fixed date, as every player got it | 6 | 100 | 148 |
| Campaign | A campaign level, chapters I to III | 6 | 100 | 165 |
| Iron | An Iron level: one life, 35% tougher waves, no abilities | 3 | 28 | 30 |
Every model gets the same boards in the same order from set v1: the daily sieges of six fixed dates, six campaign levels across the game's three chapters, and three Iron levels (one life, waves 35% tougher, no abilities). A model ranks once it has played the first 2 of every tier. After that the bench keeps playing further boards, up to 6 per tier.
A decision
Play is in paused time. Every decision is a fresh request with the whole board as text: the map with the towers drawn on it, gold, lives, every tower with its stats and upgrade costs, the attackers on the field, the next three waves (the same full intel the game shows), and what happened since the last decision, including any refused orders and why. The model answers with up to 12 orders: build, upgrade (choosing a final form at tier 3), sell, retarget, the two abilities, and calling the next wave early. The first decision is planning, with the clock stopped. After that the battle runs and stops for the next decision when the field is clear, when a life is lost, or after 20 seconds of battle. After 150 decisions the battle runs to its end without further orders.
An order the game refuses is not retried; it is counted and reported back, as the game tells a player. The one retry is for a reply with no readable orders tag. A second unreadable reply waits.
Score
Ravelin scores a won battle by its time plus 5 seconds per life lost, lower is better, and shows a par: its reference bot's score on that board. Here a won battle scores par divided by its score, never below 0.5, so beating par scores above 1. A lost battle scores half the share of the waves it held, so any win beats any loss. An Iron board takes its normal version's par, since the bot cannot hold the Iron one. The benchmark score is 100 times the mean, so 100 is winning every board at par. The 95% interval is the ratio estimator's, by the delta method. The game's bot is shown on the same boards in four settings for scale.
Cost and time
Same as the chess and sudoku benchmarks: Claude models run through Claude Code, OpenAI models through the Codex CLI and Google models through the Antigravity CLI on a subscription, with every tool switched off, and local models on our own DGX Sparks. Cost is API-equivalent: what the tokens of every request in the battle, retries included, would cost at the lab's list price. Local models have no list price, so they show n/a, not zero. Time is the wall-clock wait for the replies, including any queue on our side, so compare it within a route rather than across them. A request that fails on our side (a timeout, an outage, a usage limit) aborts the game, and an aborted game never counts against the model.
What it doesn't show
Paused time removes reaction speed on purpose: the model is asked again at every clear, every lost life and every 20 seconds, and never has to act mid-thought. These scores compare with each other and with the bot, not with a person playing the web game against the clock.