What does Mystery Game Puzzles measure?

Mystery Game Puzzles is a knowledge benchmark in the BenchGecko catalog. 60 AI models have been tested on it. Scores range from 0.9 to 82.4 out of 100.

Which model leads on Mystery Game Puzzles?

GPT-6 Astra from OpenAI leads Mystery Game Puzzles with a score of 82.4. The median score across 60 tested models is 19.6.

Is Mystery Game Puzzles saturated?

No · the top score is 82.4 out of 100 (82%). There is still meaningful room for improvement on Mystery Game Puzzles.

Does Mystery Game Puzzles predict performance on other benchmarks?

Yes · Mystery Game Puzzles scores correlate 0.95 with Osworld 2 0 across 7 shared models. Models that do well on Mystery Game Puzzles tend to do well on Osworld 2 0.

How often is Mystery Game Puzzles data refreshed?

BenchGecko pulls updates daily. New model scores on Mystery Game Puzzles appear as soon as they are published by Epoch AI or the model provider.

Benchmark · KnowledgeCompetitive

Mystery Game Puzzles

Name: Mystery Game Puzzles Benchmark
Creator: BenchGecko
License: https://creativecommons.org/licenses/by/4.0/

Updated 2026-10-13

Models tested

Top score

82.4

GPT-6 Astra

Median

19.6

min 0.9

Top-5 spread

σ 8.8

wide open

The Frontier

Best score over time · one chart, every benchmark

Chart type

Frontier on Mystery Game Puzzles rose from 21.8 to 82.4 in 17 months · +60.6 points · latest leader GPT-6 Astra from OpenAI.

Pink dots = frontier records · 7 totalClick to open model page

Full rankings

60 models tested · sorted by score

#	Model	Score	Price
1	GPT-6 Astra· OpenAI	82.4	$10.00
2	GPT-6.1 Sol· OpenAI	78.0	$2.00
3	Claude Opus 5.5· Anthropic	68.0	$4.00
4	Claude Sonnet 5.5· Anthropic	61.5	$2.00
5	Gemini 4 Argon· Google DeepMind	60.3	$2.00
6	Claude Opus 5· Anthropic	54.8	$5.00
7	Claude Fable 5.1· Anthropic	53.7	$10.00
8	GPT-5.6 Sol· OpenAI	53.7	$2.00
9	GPT-6 Sol· OpenAI	51.5	$2.00
10	Claude Fable 5· Anthropic	47.1	$10.00
11	Gemini 3.8 Flash· Google DeepMind	41.6	$0.75
12	DeepSeek V4 Pro 0813· DeepSeek	37.2	$0.66
13	DeepSeek V4.1 Flash· DeepSeek	37.2	$0.15
14	Qwen3.8 Max (0902)· Alibaba Qwen	31.7	$2.00
15	Gemini 3.7 Flash· Google DeepMind	30.6	$0.75
16	GPT-5.4· OpenAI	30.6	$2.50
17	Claude Opus 4.8· Anthropic	29.5	$5.00
18	Claude Sonnet 5· Anthropic	28.4	$2.00
19	GPT-5.6 Terra· OpenAI	28.4	$2.00
20	DeepSeek V4 Flash 0731· DeepSeek	27.3	$0.02
21	Gemini 3.1 Pro Preview· Google DeepMind	27.3	$2.00
22	Grok 4.6· xAI	27.3	$2.00
23	GLM 5.3· z-ai	26.2	$0.03
24	Gemini 3.5 Flash· Google DeepMind	25.1	$1.50
25	Qwen3.7 Max· Alibaba Qwen	25.1	$1.48
26	Claude Haiku 5.5· Anthropic	22.9	$0.10
27	Gemini 3.6 Flash· Google DeepMind	22.9	$0.75
28	Grok 4.7· xAI	21.8	$2.00
29	o3· OpenAI	21.8	$2.00
30	Claude Opus 4.7· Anthropic	20.7	$5.00
31	Gemini 3 Flash Preview· Google DeepMind	18.5	$0.50
32	Kimi K3· moonshotai	18.5	$0.45
33	Claude Opus 4.6· Anthropic	17.4	$5.00
34	GPT-5· OpenAI	15.2	$1.25
35	GPT-5.2· OpenAI	15.2	$1.75
36	Claude Opus 4.5· Anthropic	14.1	$5.00
37	Qwen3.6 35B A3B· Alibaba Qwen	14.1	$0.15
38	Claude Opus 4.1· Anthropic	13.0	$15.00
39	GPT-5.6 Luna· OpenAI	13.0	$0.20
40	Nemotron 3 Ultra· NVIDIA	11.9	$0.60
41	Qwen3.5-Flash· Alibaba Qwen	11.9	$0.07
42	Gemini 3.5 Flash Lite· Google DeepMind	10.8	$0.30
43	GLM 5.2· z-ai	10.8	$0.02
44	GPT-5.1· OpenAI	10.8	$1.25
45	Qwen3.6 Max Preview· Alibaba Qwen	10.8	$1.03
46	Kimi K2.6· moonshotai	9.7	$0.95
47	Qwen3.5 397B A17B· Alibaba Qwen	9.7	$0.55
48	Qwen3.6 Flash· Alibaba Qwen	9.7	$0.19
49	Claude Sonnet 4.5· Anthropic	8.6	$3.00
50	DeepSeek V4 Pro· DeepSeek	8.6	$0.96
51	Mistral Large 4· Mistral AI	8.6	$0.68
52	Qwen 3.5 Plus (hosted 397B-A17B)· Alibaba Qwen	8.6	n/a
53	Qwen3.7 Plus· Alibaba Qwen	8.6	$0.32
54	Claude Sonnet 4.6· Anthropic	7.5	$3.00
55	Qwen3.7 Flash· Alibaba Qwen	6.4	$0.03
56	GPT-4 Turbo· OpenAI	3.1	$10.00
57	GPT-4o-mini· OpenAI	3.1	$0.15
58	Qwen3.6 Plus· Alibaba Qwen	3.1	$0.33
59	GPT-5.4 Mini· OpenAI	2.0	$0.75
60	GPT-5 Mini· OpenAI	0.9	$0.25