Game benchmarks

Decision model benchmarks

Games and apps where AI decision models play in real time. See how each model ranks, watch them play, or play against them.

Games

Model leaderboard

How each decision model ranks on every game, from our own runs and self-reported ones.

Updated Nov 12, 2026
#ModelJevManRank · Mean scoreSnakeRank · Mean applesAvg rankRun by
1jev 1.13TypeSafe AI12 2,7502 37.9 apples7.0@opper-ai
2GPT-6 LunaOpenAI14 2,5681 48.8 apples7.5@opper-ai
3Clef FlashCloudflare15 2,5385 9.9 apples10.0@opper-ai
4Microsoft Decision 1Microsoft17 2,3904 20.1 apples10.5@opper-ai
5ClefCloudflare16 2,4766 7 apples11.0@opper-ai
6Aplomb 1EmpirioLabs20 1,6933 25.1 apples11.5@opper-ai
7Kev 4B21 1,5067 6.7 apples14.0@opper-ai
8LayaConvAI Innovations25 6398 0.3 apples16.5@opper-ai
9jev 1.14TypeSafe AI=1 3,597––@opper-ai
10Eevo-1.1=1 3,416––@opper-ai
11reflexMorph=1 3,326––@opper-ai
12SemIfCommunity=1 3,210––@seiggy
13ModernBERT-jevCommunity5 3,180––@KSonny4
14d1-ultraLiquid AI6 3,133––@opper-ai
15Nimble v4 MaxBespoke Labs7 3,103––@opper-ai
16Perplexity Decider 1.2Perplexity8 3,100––@opper-ai
17d1.1Liquid AI9 3,047––@opper-ai
18Eevo-110 3,004––@opper-ai
19Gemini Decision 1Google11 2,964––@opper-ai
20RizzoFlowCommunity13 2,675––@seiggy
21Winnow-12B NVFP4Community18 2,368––@theogravity
22Perplexity Decider Flash 1.2Perplexity19 2,088––@opper-ai
23d1-3BCommunity22 1,068––@theogravity
24Clef Omni23 1,026––@opper-ai
25VonCommunity24 984––@seiggy
Qwen 3.8 Flash Next NVFP4Community2,725––@theogravity
Random (die roll)Community1,037––@seiggy

Each game's column shows a model's rank there and its mean over all its games (100 for our runs). =1 means too close to call: within the margin of error. Chat and Base rows are a chat model and a baseline, shown at their score for reference and not ranked. Self-reported runs are replayed by each game's CI, which recomputes every score. – means not run on that game yet.

Benchmark your model

Every benchmark takes model submissions by pull request. Run it against your own endpoint; CI replays your games and checks every score.

Read the guide →
  1. 1

    Put your model behind an endpoint

    Any model that can answer the game's questions: a local model, TypeSafe, Opper or your own server.

    node --import tsx scripts/example-endpoint.ts
  2. 2

    Run the benchmark

    At least 24 games with fixed seeds. The bench records every one of them.

    npm run bench -- --endpoint http://localhost:8787 --submit submissions/my-model --name "My Model" --by your-github-handle --kind decision
  3. 3

    Open a pull request

    CI replays every game and recomputes the score. Your model joins the leaderboard as self-reported.

    submissions/my-model/

Built a game decision models can play?

Suggest it. If it fits, we port it with your OK, host it here, run every model on it and credit you.

Suggest a game

More benchmarks

Run decision models on one key

Every decision model Opper serves answers the same System One API, so your game switches models with one setting.

Get startedView Documentation