Reasoning benchmark

DTBench leaderboard

As of October 2026, Claude Opus 5.5 has the highest published DTBench score on Noometry at 98.9%, out of 151 models with results.

Last verified October 10, 2026

About DTBench

A description with primary sources is being prepared for this benchmark.

Category: Reasoning
Introduced: 2026
Unit: Percent (random guessing ≈ 40%)
Official site: epoch.ai

Top 15 models

Claude Opus 5.5 98.9%
Claude Fable 5 98.4%
Claude Opus 5 97.9%
Claude Fable 5.1 97.6%
GPT-6 Astra 97.3%
GPT-6 Sol 97.3%
Grok 4.6 97.3%
Gemini 3.1 Pro Preview 97.1%
Gemini 3.7 Flash 96.8%
Grok 4.5 96.5%
Muse Spark 1.3 96.5%
GPT-5.5 96%
GPT-5.5 Pro 96%
GPT-5.6 Sol 96%
Grok 4.7 96%
9596979899

Sponsored placements are available on pages like this one. Advertise on Noometry

All results

DTBench results by model
#	Model	Provider	Score	Setting	Source
1	Claude Opus 5.5	Anthropic	98.9%	max	Epoch AI
2	Claude Fable 5	Anthropic	98.4%	max	Epoch AI
3	Claude Opus 5	Anthropic	97.9%	high	Epoch AI
4	Claude Fable 5.1	Anthropic	97.6%		Epoch AI
5	GPT-6 Astra	OpenAI	97.3%	xhigh	Epoch AI
6	GPT-6 Sol	OpenAI	97.3%	max	Epoch AI
7	Grok 4.6	xAI	97.3%	xhigh	Epoch AI
8	Gemini 3.1 Pro Preview	Google	97.1%	medium	Epoch AI
9	Gemini 3.7 Flash	Google	96.8%	high	Epoch AI
10	Grok 4.5	xAI	96.5%	high	Epoch AI
11	Muse Spark 1.3	Meta	96.5%		Epoch AI
12	GPT-5.5	OpenAI	96%	xhigh	Epoch AI
13	GPT-5.5 Pro	OpenAI	96%	xhigh	Epoch AI
14	GPT-5.6 Sol	OpenAI	96%	promax	Epoch AI
15	Grok 4.7	xAI	96%	xhigh	Epoch AI
16	Gemini 3.8 Flash	Google	95.7%		Epoch AI
17	Gemini 3.6 Flash	Google	95.5%	high	Epoch AI
18	Claude Opus 4.8	Anthropic	94.9%	max	Epoch AI
19	Gemini 3.5 Flash	Google	94.7%	high	Epoch AI
20	Muse Spark 1.2	Meta	94.7%	xhigh	Epoch AI
21	Claude Opus 4.7	Anthropic	94.7%	max	Epoch AI
22	GPT-5.4	OpenAI	94.4%	xhigh	Epoch AI
23	Muse Spark 1.1	Meta	94.4%	high	Epoch AI
24	DeepSeek V4 Pro	DeepSeek	93.9%		Epoch AI
25	GLM-5.2	Z.ai (Zhipu)	93.6%	max	Epoch AI
26	GPT-5.6 Terra	OpenAI	93.3%	max	Epoch AI
27	Claude Sonnet 5	Anthropic	92.5%	max	Epoch AI
28	Qwen3.7 Max	Alibaba (Qwen)	92.3%		Epoch AI
29	Qwen3.8 Max	Alibaba (Qwen)	92%	xhigh	Epoch AI
30	Claude Opus 4.6	Anthropic	91.2%	max	Epoch AI
31	Kimi K3	Moonshot AI	91.2%	max	Epoch AI
32	DeepSeek V4 Flash	DeepSeek	90.9%	max	Epoch AI
33	GPT-5.2	OpenAI	90.9%	xhigh	Epoch AI
34	Kimi K2.6	Moonshot AI	90.9%		Epoch AI
35	GPT-5	OpenAI	90.7%	high	Epoch AI
36	Grok 4.3	xAI	90.7%	high	Epoch AI
37	GPT-5.1	OpenAI	90.1%	high	Epoch AI
38	GPT-6 Luna	OpenAI	90.1%	max	Epoch AI
39	Grok 4.20 (Non-Reasoning)	xAI	90.1%		Epoch AI
40	Nemotron 3 Ultra	NVIDIA	90.1%		Epoch AI
41	Claude Opus 4.5	Anthropic	89.9%	high	Epoch AI
42	Claude Sonnet 4.6	Anthropic	89.9%	max	Epoch AI
43	DeepSeek V4.1 Flash	DeepSeek	89.9%		Epoch AI
44	Gemini 3 Flash Preview	Google	89.1%	high	Epoch AI
45	GPT-5.6 Luna	OpenAI	89.1%	xhigh	Epoch AI
46	Qwen3.8 27B	Alibaba (Qwen)	88%		Epoch AI
47	DeepSeek-V3.2-Exp	DeepSeek	87.7%	thinking	Epoch AI
48	Grok 4.1 Fast	xAI	87.7%		Epoch AI
49	GLM-5.3	Z.ai (Zhipu)	87.7%		Epoch AI
50	Inkling	Thinking Machines Lab	87.5%	xhigh	Epoch AI
51	Qwen3.5 397B-A17B	Alibaba (Qwen)	87.5%		Epoch AI
52	Qwen3.6 Max Preview	Alibaba (Qwen)	87.2%		Epoch AI
53	o3-pro	OpenAI	86.9%	high	Epoch AI
54	o3	OpenAI	84.8%	high	Epoch AI
55	MiMo-V2.5-Pro	Xiaomi	84.5%		Epoch AI
56	Qwen3.5 122B-A10B	Alibaba (Qwen)	84.3%	none	Epoch AI
57	Qwen3.7 Plus	Alibaba (Qwen)	84%		Epoch AI
58	Gemini 3.5 Flash Lite	Google	83.5%	high	Epoch AI
59	Claude Sonnet 4.5	Anthropic	83.2%		Epoch AI
60	Qwen3.5-Flash	Alibaba (Qwen)	82.9%		Epoch AI
61	DeepSeek-V3.1	DeepSeek	82.7%	thinking	Epoch AI
62	Gemma 4 31B IT	Google	82.7%		Epoch AI
63	Grok 4 Fast	xAI	82.7%		Epoch AI
64	Gemini 2.5 Pro	Google	82.4%		Epoch AI
65	Qwen3.5 27B	Alibaba (Qwen)	82.4%		Epoch AI
66	Qwen3 Max	Alibaba (Qwen)	82.1%		Epoch AI
67	Qwen3.6 Plus	Alibaba (Qwen)	81.9%		Epoch AI
68	Claude Opus 4	Anthropic	81.6%		Epoch AI
69	DeepSeek-V3.1-Terminus	DeepSeek	81.3%		Epoch AI
70	Qwen Plus	Alibaba (Qwen)	81.1%		Epoch AI
71	GPT-5 Mini	OpenAI	80.5%	high	Epoch AI
72	Qwen3.5 Plus	Alibaba (Qwen)	80.5%		Epoch AI
73	GPT-5.4 nano	OpenAI	80.3%	xhigh	Epoch AI
74	Qwen3 235B-A22B	Alibaba (Qwen)	80.3%		Epoch AI
75	Claude Opus 4.1	Anthropic	80%		Epoch AI
76	GPT-5.4 mini	OpenAI	80%	xhigh	Epoch AI
77	Qwen3.5 35B-A3B	Alibaba (Qwen)	80%		Epoch AI
78	MiniMax-M3	MiniMax	78.9%		Epoch AI
79	Qwen3.6 27B	Alibaba (Qwen)	78.1%		Epoch AI
80	o4-mini	OpenAI	77.6%	high	Epoch AI
81	Claude Sonnet 4	Anthropic	77.1%		Epoch AI
82	Qwen3.6 Flash	Alibaba (Qwen)	77.1%		Epoch AI
83	Gemini 3.1 Flash Lite	Google	76.8%	high	Epoch AI
84	Gemini 2.5 Flash	Google	76.5%		Epoch AI
85	gpt-oss-120b	OpenAI	76.3%	high	Epoch AI
86	Mistral Medium	Mistral AI	75.5%		Epoch AI
87	Gemma 4 26B A4B IT	Google	74.9%		Epoch AI
88	o1	OpenAI	74.7%	high	Epoch AI
89	Qwen3.6 35B-A3B	Alibaba (Qwen)	73.9%		Epoch AI
90	Claude Haiku 4.5	Anthropic	73.6%		Epoch AI
91	Qwen3.5-9B	Alibaba (Qwen)	71.2%		Epoch AI
92	Mistral Small	Mistral AI	70.9%		Epoch AI
93	Qwen3-30B-A3B	Alibaba (Qwen)	69.3%		Epoch AI
94	GPT-4.1 mini	OpenAI	68.8%		Epoch AI
95	o3-mini	OpenAI	68.8%	high	Epoch AI
96	GPT-4.1	OpenAI	68.3%		Epoch AI
97	gpt-oss-20b	OpenAI	68%	high	Epoch AI
98	Claude 3.5 Sonnet	Anthropic	67.8%		Epoch AI
99	Qwen3 32B	Alibaba (Qwen)	67.5%		Epoch AI
100	Grok-2 (Dec 2024)	xAI	65.2%		Epoch AI
101	Mistral Large	Mistral AI	65.1%		Epoch AI
102	DeepSeek-V3	DeepSeek	64.8%		Epoch AI
103	GPT-4o	OpenAI	64.5%		Epoch AI
104	Qwen3 14B	Alibaba (Qwen)	64%		Epoch AI
105	Gemini 2.0 Flash (Feb 2025)	Google	63.2%		Epoch AI
106	Qwen2.5 72B Instruct	Alibaba (Qwen)	62.9%		Epoch AI
107	Gemini 2.5 Flash-Lite	Google	62.8%		Epoch AI
108	GPT-4	OpenAI	62.7%		Epoch AI
109	GPT-5 Nano	OpenAI	62.7%	high	Epoch AI
110	Llama 4 Maverick	Meta	61.9%		Epoch AI
111	Claude 3 Opus	Anthropic	61.6%		Epoch AI
112	GPT-4 Turbo	OpenAI	61.6%		Epoch AI
113	Llama 3.1-405B	Meta	61.4%		Epoch AI
114	Command A	Cohere	61.3%		Epoch AI
115	Magistral Small	Mistral AI	61.3%		Epoch AI
116	Llama 3.1-70B	Meta	60%		Epoch AI
117	Qwen3 8B	Alibaba (Qwen)	59.7%		Epoch AI
118	Llama-3.3-70B-Instruct	Meta	59.5%		Epoch AI
119	Gemini 1.5 Pro (May 2024)	Google	59%		Epoch AI
120	Llama 4 Scout	Meta	57.9%		Epoch AI
121	Claude 3.5 Haiku	Anthropic	56.7%		Epoch AI
122	Mixtral 8x22B	Mistral AI	55.1%		Epoch AI
123	Command R+	Cohere	54.9%		Epoch AI
124	GPT-4o mini	OpenAI	54.4%		Epoch AI
125	Llama 3-70B	Meta	54.2%		Epoch AI
126	Gemini 1.5 Flash (May 2024)	Google	53.8%		Epoch AI
127	Claude 3 Sonnet	Anthropic	53.6%		Epoch AI
128	Gemini 2.0 Flash-Lite	Google	52.5%		Epoch AI
129	Gemma 3 27B	Google	52.5%		Epoch AI
130	GPT-4.1 nano	OpenAI	52.5%		Epoch AI
131	Claude 2	Anthropic	51.9%		Epoch AI
132	Ministral 3B	Mistral AI	51.7%		Epoch AI
133	Claude 2.1	Anthropic	51%		Epoch AI
134	Gemma 3 4B	Google	50.9%		Epoch AI
135	Llama 3.1-8B	Meta	50.9%		Epoch AI
136	Claude 3 Haiku	Anthropic	50.1%		Epoch AI
137	Gemini 1.5 Flash 8B	Google	50%		Epoch AI
138	Mixtral 8x7B	Mistral AI	49.6%		Epoch AI
139	Gemma 3 12B	Google	48.8%		Epoch AI
140	Mistral Nemo	Mistral AI	48.6%		Epoch AI
141	GPT-3.5-turbo	OpenAI	48.5%		Epoch AI
142	Gemma 2 27B	Google	48%		Epoch AI
143	Qwen2.5 7B Instruct	Alibaba (Qwen)	47.7%		Epoch AI
144	Command R	Cohere	46.4%		Epoch AI
145	Gemini 1.0 Pro	Google	45.9%		Epoch AI
146	Claude Instant	Anthropic	45.8%		Epoch AI
147	Ministral 8B	Mistral AI	45.7%		Epoch AI
148	Llama 3-8B	Meta	43.9%		Epoch AI
149	Mistral 7B	Mistral AI	42.5%		Epoch AI
150	Llama 2-13B	Meta	42.2%		Epoch AI
151	Llama 2-70B	Meta	41.6%		Epoch AI

Compare the leaders

Other reasoning benchmarks

Frequently asked questions

Which model has the highest DTBench score?

As of October 2026, Claude Opus 5.5 has the highest published DTBench score on Noometry at 98.9%, out of 151 models with results.

What is the best open-weight model on DTBench?

DeepSeek V4 Pro has the highest DTBench accuracy among open-weight models at 93.9%, ranking 24 of 151 overall.