What does Dtbench measure?

Dtbench is a knowledge benchmark in the BenchGecko catalog. 132 AI models have been tested on it. Scores range from 3.7 to 98.2 out of 100.

Which model leads on Dtbench?

Claude Opus 5.5 from Anthropic leads Dtbench with a score of 98.2. The median score across 132 tested models is 64.4.

Is Dtbench saturated?

Yes · the top model on Dtbench has reached 98.2 out of 100, within 5% of the theoretical ceiling. This benchmark is approaching saturation and may be replaced by a harder successor.

Does Dtbench predict performance on other benchmarks?

Yes · Dtbench scores correlate 0.98 with OpenCompass · GPQA-Diamond across 11 shared models. Models that do well on Dtbench tend to do well on OpenCompass · GPQA-Diamond.

How often is Dtbench data refreshed?

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

Benchmark · KnowledgeSettled

Dtbench

Name: Dtbench Benchmark
Creator: BenchGecko
License: https://creativecommons.org/licenses/by/4.0/

Updated 2026-09-22

Models tested

132

Top score

98.2

Claude Opus 5.5

Median

64.4

min 3.7

Top-5 spread

σ 1.0

Settled

The Frontier

Best score over time · one chart, every benchmark

Chart type

Frontier on Dtbench rose from 40.9 to 98.2 in 22 months · +57.3 points · latest leader Claude Opus 5.5 from Anthropic.

Pink dots = frontier records · 10 totalClick to open model page

Full rankings

132 models tested · sorted by score

#	Model	Score	Price
1	Claude Opus 5.5· Anthropic	98.2	$4.00
2	Claude Fable 5· Anthropic	97.3	$10.00
3	Claude Opus 5· Anthropic	96.5	$5.00
4	Claude Fable 5.1· Anthropic	96.0	$10.00
5	GPT-6 Astra· OpenAI	95.5	$10.00
6	Grok 4.6· xAI	95.5	$2.00
7	Gemini 3.1 Pro Preview· Google DeepMind	95.1	$2.00
8	Gemini 3.7 Flash· Google DeepMind	94.7	$0.75
9	Grok 4.5· xAI	94.2	$2.00
10	GPT-5.6 Sol· OpenAI	93.3	$2.00
11	Gemini 3.8 Flash· Google DeepMind	92.9	$0.75
12	Gemini 3.6 Flash· Google DeepMind	92.5	$0.75
13	Claude Opus 4.8· Anthropic	91.5	$5.00
14	Gemini 3.5 Flash· Google DeepMind	91.1	$1.50
15	Claude Opus 4.7· Anthropic	91.1	$5.00
16	GPT-5.4· OpenAI	90.7	$2.50
17	DeepSeek V4 Pro 0813· DeepSeek	89.8	$0.66
18	GLM 5.2· z-ai	89.3	$1.00
19	GPT-5.6 Terra· OpenAI	88.9	$2.00
20	Claude Sonnet 5· Anthropic	87.5	$2.00
21	Qwen3.7 Max· Alibaba Qwen	87.1	$1.25
22	Qwen3.8 Max (0902)· Alibaba Qwen	86.7	$2.00
23	Claude Opus 4.6· Anthropic	85.3	$5.00
24	Kimi K3· moonshotai	85.3	$2.70
25	DeepSeek V4 Flash 0731· DeepSeek	84.9	$0.01
26	GPT-5.2· OpenAI	84.9	$1.75
27	Kimi K2.6· moonshotai	84.9	$0.95
28	DeepSeek V4 Pro· DeepSeek	84.5	$0.21
29	GPT-5· OpenAI	84.5	$1.25
30	Grok 4.3· xAI	84.5	$1.25
31	GPT-5.1· OpenAI	83.5	$1.25
32	Grok 4.20· xAI	83.5	$1.25
33	Nemotron 3 Ultra· NVIDIA	83.5	$0.50
34	Claude Opus 4.5· Anthropic	83.1	$5.00
35	Claude Sonnet 4.6· Anthropic	83.1	$3.00
36	DeepSeek V4.1 Flash· DeepSeek	83.1	$0.15
37	Gemini 3 Flash Preview· Google DeepMind	81.8	$0.50
38	GPT-5.6 Luna· OpenAI	81.8	$0.20
39	Qwen3.8 27B· Alibaba Qwen	80.0	$0.42
40	DeepSeek V3.2 Exp· DeepSeek	79.5	$0.27
41	GLM 5.3· z-ai	79.5	$1.40
42	Qwen3.5 397B A17B· Alibaba Qwen	79.1	$0.55
43	Qwen3.6 Max Preview· Alibaba Qwen	78.7	$1.03
44	o3 Pro· OpenAI	78.2	$20.00
45	DeepSeek V4 Flash· DeepSeek	77.3	$0.03
46	DeepSeek V3.2· DeepSeek	76.0	$0.28
47	o3· OpenAI	74.7	$2.00
48	Qwen3.7 Plus· Alibaba Qwen	73.3	$0.32
49	Gemini 3.5 Flash Lite· Google DeepMind	72.5	$0.30
50	Claude Sonnet 4.5· Anthropic	72.0	$3.00
51	Qwen3.5-Flash· Alibaba Qwen	71.5	$0.07
52	DeepSeek V3.1· DeepSeek	71.1	$0.25
53	Gemma 4 31B· Google DeepMind	71.1	$0.09
54	Grok 4 Fast· xAI	71.1	—
55	Gemini 2.5 Pro· Google DeepMind	70.7	$1.25
56	Qwen3 Max· Alibaba Qwen	70.2	$0.78
57	Qwen3.6 Plus· Alibaba Qwen	69.8	$0.33
58	Claude Opus 4· Anthropic	69.3	$15.00
59	GPT-5 Mini· OpenAI	67.5	$0.25
60	Qwen 3.5 Plus (hosted 397B-A17B)· Alibaba Qwen	67.5	—
61	GPT-5.4 Nano· OpenAI	67.1	$0.20
62	Qwen3 235B A22B Thinking 2507· Alibaba Qwen	67.1	$0.23
63	Claude Opus 4.1· Anthropic	66.7	$15.00
64	GPT-5.4 Mini· OpenAI	66.7	$0.75
65	Qwen3.5-35B-A3B· Alibaba Qwen	66.7	$0.15
66	MiniMax M3· minimax	64.9	$0.30
67	Qwen3 235B A22B Instruct 2507· Alibaba Qwen	64.0	$0.09
68	Qwen3.6 27B· Alibaba Qwen	63.5	$0.32
69	o4 Mini· OpenAI	62.7	$1.10
70	Claude Sonnet 4· Anthropic	61.8	$3.00
71	Qwen3.6 Flash· Alibaba Qwen	61.8	$0.19
72	Gemini 3.1 Flash Lite· Google DeepMind	61.3	$0.25
73	Gemini 2.5 Flash· Google DeepMind	60.9	$0.30
74	gpt-oss-120b· OpenAI	60.5	$0.04
75	Qwen3 235B A22B· Alibaba Qwen	59.5	$0.46
76	Mistral Medium 3.5· Mistral AI	59.1	$1.50
77	Gemma 4 26B A4B · Google DeepMind	58.2	$0.07
78	o1· OpenAI	57.8	$15.00
79	Qwen3.6 35B A3B· Alibaba Qwen	56.5	$0.15
80	Claude Haiku 4.5· Anthropic	56.0	$1.00
81	Qwen3.5-9B· Alibaba Qwen	52.0	$0.10
82	GPT-4.1 Mini· OpenAI	48.0	$0.40
83	o3 Mini· OpenAI	48.0	$1.10
84	GPT-4.1· OpenAI	47.1	$2.00
85	gpt-oss-20b· OpenAI	46.7	$0.02
86	Claude 3.5 Sonnet· Anthropic	46.3	—
87	Qwen3 32B· Alibaba Qwen	45.8	$0.08
88	Grok-2 (Dec 2024)· xAI	42.0	—
89	DeepSeek V3· DeepSeek	41.3	$0.26
90	GPT-4o (2024-11-20)· OpenAI	40.9	$2.50
91	Qwen3 14B· Alibaba Qwen	40.0	$0.12
92	Gemini 2.0 Flash (Dec 2024)· Google DeepMind	38.7	—
93	Qwen2.5 72B Instruct· Alibaba Qwen	38.2	$0.36
94	Gemini 2.5 Flash Lite· Google DeepMind	38.0	$0.10
95	GPT-4 Turbo· OpenAI	37.8	$10.00
96	GPT-5 Nano· OpenAI	37.8	$0.05
97	Mistral Medium 3· Mistral AI	37.1	$0.40
98	Llama 4 Maverick· Meta	36.5	$0.19
99	Claude 3 Opus· Anthropic	36.0	—
100	Llama 3.1 405B· Meta	35.6	—
101	Qwen3 30B A3B· Alibaba Qwen	33.8	$0.12
102	Llama 3.1 70B Instruct· Meta	33.3	$0.40
103	Mistral Small 3.2 24B· Mistral AI	33.1	$0.09
104	Qwen3 8B· Alibaba Qwen	32.9	$0.12
105	Llama 3.3 70B Instruct· Meta	32.5	$0.10
106	Mistral Small 3.1 24B· Mistral AI	31.1	$0.35
107	Llama 4 Scout· Meta	29.8	$0.10
108	Claude 3.5 Haiku· Anthropic	27.8	—
109	Gemini 1.5 Pro (May 2024)· Google DeepMind	27.4	—
110	Mistral Large· Mistral AI	26.5	$2.00
111	Mixtral 8x22B Instruct· Mistral AI	25.2	$2.00
112	Command R (08-2024)· Cohere	24.9	$0.15
113	GPT-4o-mini· OpenAI	24.0	$0.15
114	Llama 3-70B· Meta	23.6	—
115	Claude 3 Sonnet· Anthropic	22.7	—
116	Gemma 3 27B· Google DeepMind	20.9	$0.08
117	GPT-4.1 Nano· OpenAI	20.9	$0.10
118	Claude 2· Anthropic	19.8	—
119	Gemini 1.5 Flash (May 2024)· Google DeepMind	19.0	—
120	Claude 2.1· Anthropic	18.3	—
121	Gemma 3 4B· Google DeepMind	18.2	$0.05
122	Llama 3.1 8B Instruct· Meta	18.2	$0.02
123	Claude 3 Haiku· Anthropic	16.9	$0.25
124	Mixtral 8x7B· Mistral AI	16.0	—
125	Gemma 3 12B· Google DeepMind	14.7	$0.05
126	Mistral Nemo· Mistral AI	14.3	$0.02
127	GPT-3.5 Turbo (older v0613)· OpenAI	14.2	$1.00
128	Gemma 2 27B· Google DeepMind	13.3	$0.65
129	Qwen2.5 7B Instruct· Alibaba Qwen	12.9	$0.10
130	Gemini 1.0 Pro· Google DeepMind	9.8	—
131	Claude Instant· Anthropic	9.6	—
132	Llama 2-13B· Meta	3.7	—