SShogo Bench

DTBench

Decision-theory questions.

Agent Index weight
0%
shown for context only
Models tested
172
from the benchmark’s leaderboard and reports
Normalised for the index
0.4 → 0, 1 → 100
Official site

Results

#ModelScoreNormalisedHarnessCost/taskDate
1Claude Fable 5.2 Supported
Anthropic
99.2%source——
2Claude Opus 5.5 Supported
Anthropic
98.9%source——
3Claude Fable 5 Supported
Anthropic
98.4%source——
4Claude Opus 5 Supported
Anthropic
97.9%source——
5Claude Fable 5.1 Supported
Anthropic
97.6%source——
6Gemini 4 Argon Supported
Google
97.6%source——
7Claude Sonnet 5.5 Estimated
Anthropic
97.3%source——
8GPT-6 Astra Supported
OpenAI
97.3%source——
9GPT-6 Sol Estimated
OpenAI
97.3%source——
10Grok 4.6 Supported
xAI
97.3%source——
11Gemini 3.1 Pro Preview Supported
Google
97.1%source——
12Gemini 3.7 Flash Estimated
Google
96.8%source——
13GPT-6.1 Sol Estimated
OpenAI
96.5%source——
14Grok 4.5 Estimated
xAI
96.5%source——
15Muse Spark 1.3 Estimated
Meta
96.5%source——
16GPT-5.5 Supported
OpenAI
96.0%source——
17GPT-5.5 Pro
OpenAI
96.0%source——
18GPT-5.6 Sol Supported
OpenAI
96.0%source——
19Grok 4.7 Estimated
xAI
96.0%source——
20Gemini 3.8 Flash Supported
Google
95.7%source——
21Gemini 3.6 Flash Estimated
Google
95.5%source——
22Muse Wonder Estimated
Meta
95.2%source——
23Claude Haiku 5.5 Estimated
Anthropic
95.0%source——
24Claude Opus 4.8 Supported
Anthropic
94.9%source——
25Gemini 3.5 Flash Supported
Google
94.7%source——
26Muse Spark 1.2 Estimated
Meta
94.7%source——
27Claude Opus 4.7 Supported
Anthropic
94.7%source——
28GPT-5.4 Supported
OpenAI
94.4%source——
29Muse Spark 1.1 Estimated
Meta
94.4%source——
30DeepSeek V4 Pro 0813
DeepSeek
93.9%source——
31GLM-5.2 Supported
Z.ai (Zhipu)
93.6%source——
32GPT-5.6 Terra Estimated
OpenAI
93.3%source——
33Kimi K3.1 Pro
Moonshot AI
93.3%source——
34Claude Sonnet 5 Supported
Anthropic
92.5%source——
35Qwen3.7 Max
Alibaba (Qwen)
92.3%source——
36Qwen3.8 Max Preview Estimated
Alibaba (Qwen)
92.0%source——
37Claude Opus 4.6 Supported
Anthropic
91.2%source——
38Kimi K3 Supported
Moonshot AI
91.2%source——
39DeepSeek V4 Flash 0731
DeepSeek
90.9%source——
40GPT-5.2 Supported
OpenAI
90.9%source——
41Kimi K2.6 Estimated
Moonshot AI
90.9%source——
42DeepSeek V4 Pro Estimated
DeepSeek
90.7%source——
43GPT-5 Supported
OpenAI
90.7%source——
44Grok 4.3 Beta
xAI
90.7%source——
45GPT-5.1 Supported
OpenAI
90.1%source——
46GPT-6 Luna
OpenAI
90.1%source——
47Grok 4.20 (Non-Reasoning) Estimated
xAI
90.1%source——
48Nemotron 3 Ultra
NVIDIA
90.1%source——
49Claude Opus 4.5 (latest) Supported
Anthropic
89.9%source——
50Claude Sonnet 4.6 Supported
Anthropic
89.9%source——
51DeepSeek V4.1 Flash Estimated
DeepSeek
89.9%source——
52Gemini 3 Flash Preview Supported
Google
89.1%source——
53GPT-5.6 Luna Estimated
OpenAI
89.1%source——
54Mistral Large 4
Mistral
88.2%source——
55Qwen3.8 27B
Alibaba (Qwen)
88.0%source——
56DeepSeek V3.2 Supported
DeepSeek
87.7%source——
57Grok 4.1 Fast (Non-Reasoning)
xAI
87.7%source——
58GLM-5.3 Supported
Z.ai (Zhipu)
87.7%source——
59Inkling Estimated
thinkingmachines
87.5%source——
60Qwen3.5 397B-A17B
Alibaba (Qwen)
87.5%source——
61Qwen3.6 Max Preview Estimated
Alibaba (Qwen)
87.2%source——
62o3-pro
OpenAI
86.9%source——
63DeepSeek V4 Flash
DeepSeek
86.4%source——
64o3 Supported
OpenAI
84.8%source——
65MiMo-V2.5-Pro
Xiaomi
84.5%source——
66Qwen3.5 122B-A10B
Alibaba (Qwen)
84.3%source——
67Qwen3.7 Plus
Alibaba (Qwen)
84.0%source——
68Gemini 3.5 Flash Lite
Google
83.5%source——
69Claude Sonnet 4.5 (latest) Supported
Anthropic
83.2%source——
70Qwen3.5 Flash
Alibaba (Qwen)
82.9%source——
71DeepSeek V3.1
DeepSeek
82.7%source——
72Gemma 4 31B IT
Google
82.7%source——
73Grok 4 Fast
xAI
82.7%source——
74Gemini 2.5 Pro Supported
Google
82.4%source——
75Qwen3.5 27B
Alibaba (Qwen)
82.4%source——
76Qwen3 Max Estimated
Alibaba (Qwen)
82.1%source——
77Qwen3.6 Plus Estimated
Alibaba (Qwen)
81.9%source——
78Claude-Opus-4 Supported
Anthropic
81.6%source——
79DeepSeek V3.1 Terminus
DeepSeek
81.3%source——
80Qwen Plus
Alibaba (Qwen)
81.1%source——
81GPT-5 Mini Estimated
OpenAI
80.5%source——
82Qwen3.5 Plus Estimated
Alibaba (Qwen)
80.5%source——
83GPT-5.4 nano
OpenAI
80.3%source——
84Qwen3 235B A22B Thinking 2507
Alibaba (Qwen)
80.3%source——
85Claude Opus 4.1 (latest) Supported
Anthropic
80.0%source——
86GPT-5.4 mini
OpenAI
80.0%source——
87Qwen3.5 35B-A3B
Alibaba (Qwen)
80.0%source——
88MiniMax-M3 Estimated
MiniMax
78.9%source——
89Qwen3 235B A22B Instruct 2507
Alibaba (Qwen)
78.4%source——
90Qwen3.6 27B
Alibaba (Qwen)
78.1%source——
91o4-mini Estimated
OpenAI
77.6%source——
92Claude Sonnet 4 Supported
Anthropic
77.1%source——
93Qwen3.6 Flash
Alibaba (Qwen)
77.1%source——
94Gemini 3.1 Flash Lite
Google
76.8%source——
95Gemini 2.5 Flash Estimated
Google
76.5%source——
96GPT OSS 120B Supported
OpenAI
76.3%source——
97Qwen3 235B A22B
Alibaba (Qwen)
75.7%source——
98Mistral Medium 3.5
Mistral
75.5%source——
99Gemma 4 26B A4B
Google
74.9%source——
100o1 Estimated
OpenAI
74.7%source——
101Qwen3.6 35B-A3B
Alibaba (Qwen)
73.9%source——
102Claude Haiku 4.5 Estimated
Anthropic
73.6%source——
103Qwen3.5 9B
Alibaba (Qwen)
71.2%source——
104Mistral Small 4
Mistral
70.9%source——
105Qwen3-30B-A3B-Thinking (Jul 2025)
Alibaba (Qwen)
69.3%source——
106GPT-4.1 mini
OpenAI
68.8%source——
107o3-mini Estimated
OpenAI
68.8%source——
108GPT-4.1
OpenAI
68.3%source——
109GPT OSS 20B
OpenAI
68.0%source——
110Claude-Sonnet-3.5 Estimated
Anthropic
67.8%source——
111Mistral Medium 3.1
Mistral
67.5%source——
112Qwen3 32B
Alibaba (Qwen)
67.5%source——
113Qwen3 30B A3B Instruct 2507
Alibaba (Qwen)
67.2%source——
114Grok-2 (Dec 2024)
xAI
65.2%source——
115Mistral Large 3
Mistral
65.1%source——
116DeepSeek V3 0324
DeepSeek
64.8%source——
117GPT-4o (2024-08-06) Supported
OpenAI
64.5%source——
118Qwen3 14B
Alibaba (Qwen)
64.0%source——
119Gemini-2.0-Flash
Google
63.2%source——
120Qwen2.5-72B Estimated
Alibaba (Qwen)
62.9%source——
121Gemini 2.5 Flash-Lite
Google
62.8%source——
122GPT-4
OpenAI
62.7%source——
123GPT-5 Nano
OpenAI
62.7%source——
124Mistral Medium 3
Mistral
62.3%source——
125Llama 4 Maverick
Meta
61.9%source——
126Claude 3 Opus Estimated
Anthropic
61.6%source——
127GPT-4 Turbo
OpenAI
61.6%source——
128Llama 3.1-405B
Meta
61.4%source——
129Cohere Command A
Cohere
61.3%source——
130Magistral Small 1.2
Mistral
61.3%source——
131Mistral Large 2 (Nov 2024)
Mistral
61.2%source——
132Qwen3 30B A3B
Alibaba (Qwen)
60.3%source——
133Llama 3.1-70B
Meta
60.0%source——
134Mistral Small 3.2
Mistral
59.9%source——
135Qwen3 8B
Alibaba (Qwen)
59.7%source——
136Llama 3.3 70B
Meta
59.5%source——
137Gemini 1.5 Pro (Feb 2024)
Google
59.0%source——
138Mistral Small 3.1
Mistral
58.6%source——
139Llama 4 Scout
Meta
57.9%source——
140Claude 3.5 Haiku
Anthropic
56.7%source——
141Mistral Large
Mistral
55.9%source——
142Mixtral 8x22B
Mistral
55.1%source——
143Command R+
Cohere
54.9%source——
144GPT-4o mini
OpenAI
54.4%source——
145Llama 3-70B
Meta
54.2%source——
146Gemini 1.5 Flash (Sep 2024)
Google
53.8%source——
147Claude 3 Sonnet
Anthropic
53.6%source——
148Gemini-2.0-Flash-Lite
Google
52.5%source——
149Gemma 3 27B
Google
52.5%source——
150GPT-4.1 nano
OpenAI
52.5%source——
151Claude 2
Anthropic
51.9%source——
152Ministral 3B
Mistral
51.7%source——
153Claude 2.1
Anthropic
51.0%source——
154Gemma 3 4B
Google
50.9%source——
155Llama 3.1-8B
Meta
50.9%source——
156Claude-Haiku-3
Anthropic
50.1%source——
157Gemini 1.5 Flash 8B
Google
50.0%source——
158Mixtral 8x7B
Mistral
49.6%source——
159Mistral Small v24.09
Mistral
49.5%source——
160Gemma 3 12B
Google
48.8%source——
161Mistral Nemo
Mistral
48.6%source——
162GPT-3.5-turbo
OpenAI
48.5%source——
163Gemma 2 27B
Google
48.0%source——
164Qwen2.5-7B
Alibaba (Qwen)
47.7%source——
165c4ai-command-r-08-2024
Unknown
46.4%source——
166Gemini 1.0 Pro
Google
45.9%source——
167Claude Instant
Anthropic
45.8%source——
168Ministral 3 8B 2512
Mistral
45.7%source——
169Llama 3-8B
Meta
43.9%source——
170Mistral 7B v0.3
Mistral
42.5%source——
171Llama 2-13B
Meta
42.2%source——
172Llama 2-70B
Meta
41.6%source——