Reasoning & KnowledgeStill discriminating

Mystery Game Puzzles

Selecting the best next move in a deliberately undisclosed game.

  • 84 models have a score, from Epoch AI.
  • The leader is claude-fable-5-2_max at 90%, dated Nov 6, 2026.
  • The top 10 are within 34 points of each other.
  • Its newest result is dated Nov 6, 2026, and 6 results were published in the last 30 days.
  • A score on this page last changed on Nov 11, 2026.

Metric

accuracy (%)

Status

Still discriminating

Category

Reasoning & Knowledge

Maintainer

Epoch AI

Chance baseline

9.22%

What it tests

Planning from a textual game state without relying on a publicly known evaluation identity. Epoch withholds the game's name to reduce targeted preparation for the benchmark.

How it works

The suite contains 100 programmatically generated positions, each filtered to have one best next move. A minimal agent submits its move through a tool. The normalized move must match the answer key; illegal or missing moves earn no credit.

How to read a score

Higher is better. Epoch's metadata gives a 9.22% random baseline. This is move-selection accuracy, not full-game win rate.

Limitations

The undisclosed game, prompts, and transcripts make independent auditing harder. A small sample and a single game distribution limit conclusions about general reasoning or broad game-playing skill.

Leaderboard

Checked Nov 17, 2026, 07:00 UTC · scores last changed Nov 11, 2026

Best Mystery Game Puzzles score per model, ranked. Each score links to its source.

#ModelScoreSource
1Anthropic: Claude Fable 5.2(max)Anthropic90%epoch
2OpenAI: GPT-6 Astra(max)OpenAI84%epoch
3OpenAI: GPT-6.1 Sol(max)OpenAI80%epoch
4OpenAI: GPT-6.1 Terra(max)OpenAI74%epoch
5Google: Gemini 4 Carbon(high)Google DeepMind73%epoch
6Anthropic: Claude Opus 5.5(max)Anthropic71%epoch
7Anthropic: Claude Sonnet 5.5(max)Anthropic65%epoch
8Google: Gemini 4 Argon(high)Google DeepMind64%epoch
9Anthropic: Claude Fable 5.1(max)Anthropic58%epoch
10OpenAI: GPT-6 Sol(max)OpenAI56%epoch
11Anthropic: Claude Fable 5(max)Anthropic52%epoch
12muse-wonder-1.0_max(max)Meta AI51%epoch
13DeepSeek: DeepSeek V4.1 Flash 1017(max)DeepSeek49%epoch
14Google: Gemini 3.8 Flash(high)Google DeepMind47%epoch
15DeepSeek: DeepSeek V4 Pro 0813(max)DeepSeek43%epoch
16DeepSeek: DeepSeek V4.1 Flash(max)DeepSeek43%epoch
17kimi-k3.1-pro-together_max(max)Moonshot42%epoch
18qwen3.8-max_xhigh(xhigh)Alibaba38%epoch
19Google: Gemini 3.7 Flash(high)Google DeepMind37%epoch
20Anthropic: Claude Opus 5Anthropic37%epoch
21SpaceXAI: Grok 4.6(xhigh)xAI34%epoch
22DeepSeek: DeepSeek V4 Flash 0731(max)DeepSeek34%epoch
23Z.ai: GLM 5.3(max)Z.ai (Zhipu AI)33%epoch
24Google: Gemini 3.1 Pro Preview(medium)Google DeepMind32%epoch
25Anthropic: Claude Opus 4.8(xhigh)Anthropic31%epoch
26Anthropic: Claude Haiku 5.5(max)Anthropic30%epoch
27OpenAI: o3(high)OpenAI29%epoch
28SpaceXAI: Grok 4.7(xhigh)xAI29%epoch
29Google: Gemini 3.5 Flash(low)Google DeepMind28%epoch
30OpenAI: GPT-5.5(low)OpenAI28%epoch
31OpenAI: GPT-5.6 Sol(low)OpenAI26%epoch
32Qwen: Qwen3.7 Max(none)Alibaba26%epoch
33Google: Gemini 3.6 Flash(minimal)Google DeepMind25%epoch
34Google: Gemini 3 Flash Preview(minimal)Google DeepMind25%epoch
35Meta: Muse Spark 1.3(max)Meta AI25%epoch
36Qwen: Qwen3.6 35B A3B(none)Alibaba22%epoch
37claude-opus-4-5-20251101_48K(budget:48k)Anthropic22%epoch
38OpenAI: GPT-5.2(medium)OpenAI22%epoch
39Anthropic: Claude Opus 4.1(budget:24k)Anthropic21%epoch
40NVIDIA: Nemotron 3 UltraNvidia20%epoch
41OpenAI: GPT-5.6 Luna(none)OpenAI20%epoch
42qwen3.5-flashAlibaba20%epoch
43mistral-large-4-0-oss_high(high)Mistral AI20%epoch
44qwen3.6-max-preview_none(none)Alibaba19%epoch
45qwen3.6-max-preview_none(none)Alibaba19%epoch
46Qwen: Qwen3.5 397B A17B(none)Alibaba18%epoch
47qwen3.5-plus_none(none)Alibaba17%epoch
48Qwen: Qwen3.5-122B-A10B(none)Alibaba17%epoch
49Anthropic: Claude Sonnet 4.5(budget:48k)Anthropic17%epoch
50DeepSeek: DeepSeek V4 Pro 0423(none)DeepSeek17%epoch
51Mistral: Mistral Large 4(high)Mistral AI17%epoch
52Qwen: Qwen3.7 PlusAlibaba17%epoch
53OpenAI: GPT-5.1(medium)OpenAI16%epoch
54Anthropic: Claude Sonnet 5Anthropic16%epoch
55OpenAI: GPT-5.4(none)OpenAI16%epoch
56Z.ai: GLM 5.2(medium)Z.ai (Zhipu AI)15%epoch
57Anthropic: Claude Opus 4.6Anthropic15%epoch
58Qwen: Qwen3.7 FlashAlibaba15%epoch
59OpenAI: GPT-5(medium)OpenAI15%epoch
60Anthropic: Claude Sonnet 4.6Anthropic14%epoch
61Anthropic: Claude Opus 4.7Anthropic13%epoch
62gpt-4-0613OpenAI12%epoch
63Google: Gemini 3.5 Flash Lite(minimal)Google DeepMind12%epoch
64MoonshotAI: Kimi K2.6(none)Moonshot12%epoch
65MoonshotAI: Kimi K3(low)Moonshot12%epoch
66OpenAI: GPT-4o-mini (2024-07-18)OpenAI12%epoch
67Qwen: Qwen3.6 FlashAlibaba12%epoch
68Qwen: Qwen3.6 Plus(none)Alibaba12%epoch
69OpenAI: GPT-5.6 Terra(medium)OpenAI10%epoch
70Qwen: Qwen3 235B A22B Thinking 2507Alibaba9%epoch
71OpenAI: GPT-5 Nano(high)OpenAI8%epoch
72Z.ai: GLM 5.3 Flash(max)Z.ai (Zhipu AI)8%epoch
73o3-mini-2025-01-31_high(high)OpenAI7%epoch
74Qwen: Qwen3.6 27B(none)Alibaba7%epoch
75o3-mini-2025-01-31_high(high)OpenAI7%epoch
76OpenAI: GPT-6 Luna(max)OpenAI7%epoch
77OpenAI: GPT-5.4 Mini(medium)OpenAI7%epoch
78OpenAI: GPT-4.1 MiniOpenAI7%epoch
79MiniMax: MiniMax M3MiniMax7%epoch
80Thinking Machines: Inkling Small(xhigh)Thinking Machines6%epoch
81o4-mini-2025-04-16_high(high)OpenAI5%epoch
82OpenAI: GPT-5 Mini(high)OpenAI5%epoch
83qwen3-max-2025-09-23Alibaba5%epoch
84o4-mini-2025-04-16_high(high)OpenAI5%epoch
85OpenAI: GPT-5.4 Nano(high)OpenAI5%epoch
86gpt-3.5-turbo-0125OpenAI3%epoch
87OpenAI: gpt-oss-120b(high)OpenAI0%epoch

Run your own prompt on the top 3

Claude Fable 5.2, GPT-6 Astra, GPT-6.1 Sol, side by side on the same prompt, with each one's time and cost. A free account includes starter credit.

Compare them

Compare the leaders

More Reasoning & Knowledge benchmarks

GPQA Diamond

Graduate-level science questions that are hard even with a search engine.

Humanity's Last Exam

A broad, extremely hard closed-ended exam at the frontier of human knowledge.

MMLU

The classic 57-subject knowledge exam, now largely solved by frontier models.

ARC-AGI-2

Novel visual-grid puzzles built to resist memorization and brute-force scaling.

ARC-AGI-1

The original abstraction-and-reasoning grid puzzles, now solved cost-efficiently.

BIG-Bench Hard

23 reasoning tasks that once beat models, now largely saturated.

SimpleQA Verified

Short factual questions that probe hallucination and calibration.

HellaSwag

Commonsense sentence completion that is trivial for people and long resisted models.

WinoGrande

Large-scale Winograd pronoun resolution that turns on one commonsense word.

PIQA

Physical commonsense: which everyday method actually works?

OpenBookQA

Grade-school science questions needing a known fact plus commonsense to combine it.

TriviaQA

Open-domain factual recall scored by strict exact match.

ANLI

Adversarially-collected natural-language inference across three rounds.

SuperGLUE

The harder successor to GLUE, a language-understanding suite now surpassed by LLMs.

LAMBADA

Predict the final word of a passage that requires the whole context.

SimpleBench

Trick questions where ordinary people comfortably beat frontier models.

MMLU-Pro

Academic and professional reasoning across a broad set of subjects with harder multiple-choice distractors than MMLU.

GPQA Main

Graduate-level science reasoning on the main GPQA question set.

NarrativeQA

Reading comprehension and answering questions about narratives.

Natural Questions (open-book)

Finding an answer to a naturally occurring search question within supplied Wikipedia evidence.

Natural Questions (closed-book)

Recalling factual answers to naturally occurring search questions without supplied Wikipedia evidence.

MedQA

Medical knowledge and reasoning on professional medical board-examination questions.

CLEAR

Identifying whether a patient note supports, rejects, or leaves uncertain the presence of a medical condition.

MEDEC

Detecting whether a clinical narrative contains a medical error.

HEAD-QA

Biomedical knowledge and reasoning on multiple-choice examination questions.

Medbullets

Applying medical knowledge to patient scenarios in USMLE-style questions.

MedMCQA

Medical entrance-examination knowledge across clinical and foundational subjects.

PubMedQA

Interpreting biomedical abstracts to answer research questions with yes, no, or maybe.

n2c2 Clinical Trial Matching

Determining whether a patient's clinical notes satisfy clinical-trial eligibility criteria.

MedHallu

Detecting unsupported statements in biomedical question-answer pairs.

AI2 ARC Challenge

Multiple-choice school science questions selected to resist simple retrieval.

Chess Puzzles

Finding the uniquely best next move from a chess position described in text.

DTBench

Reasoning correctly about decision problems involving predictions and correlated agents.

LMCA

Judging conceptual arguments in agreement with expert human ratings.

Mystery Game Puzzles is maintained by Epoch AI. Scores on this page are collected from openly-licensed sources and each links to its origin. See the full matrix to compare every model across every benchmark.