Knowledge & reasoning

HLE w/o tools leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on HLE w/o tools is 64.4% by Claude Opus 5.5. 200 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5.5	64.4%	measured
2	Gemini 4 Argon	61.3%	estimated ± 5.7 pp, medium confidence
3	Claude Fable 5.1	60.9%	measured
4	Claude Mythos 5	59.0%	measured
5	Claude Sonnet 5.5	56.9%	measured
6	Claude Opus 5	56.3%	measured
7	Claude Fable 5	54.9%	estimated ± 5.1 pp, medium confidence
8	GPT-6.1 Sol	52.7%	estimated ± 5.0 pp, low confidence
9	Gemini 3.8 Flash	52.5%	estimated ± 5.0 pp, medium confidence
10	Muse Spark 1.1	52.2%	measured
11	GPT-6 Sol	52.1%	estimated ± 5.0 pp, medium confidence
12	Gemini 3.7 Flash	51.9%	estimated ± 5.1 pp, medium confidence
13	Grok 4.6	51.4%	estimated ± 5.0 pp, medium confidence
14	Claude Opus 4.7	51.3%	estimated ± 5.1 pp, medium confidence
15	MiMo-V2.6-Pro	50.4%	estimated ± 5.7 pp, medium confidence
16	Sakana Fugu-Ultra	50.0%	measured
17	Claude Opus 4.8	49.8%	measured
18	Muse Spark 1.3	49.4%	estimated ± 5.7 pp, medium confidence
19	Gemini 3.6 Flash	49.1%	estimated ± 5.1 pp, medium confidence
20	Pareto 26.9	49.0%	measured
21	GPT-6 Astra	48.5%	estimated ± 4.4 pp, medium confidence
22	Sakana Fugu	47.2%	measured
23	Claude Opus 4.7 (Adaptive)	46.9%	measured
24	GPT-5.6 Sol	46.6%	estimated ± 4.4 pp, high confidence
25	Gemini 3 Flash	46.1%	estimated ± 5.1 pp, medium confidence
26	Claude Haiku 5.5	45.9%	measured
27	Apodex 1.1	45.7%	estimated ± 5.3 pp, medium confidence
28	Gemini 3.1 Pro	45.4%	measured
29	Step 5 Preview	44.7%	estimated ± 4.4 pp, high confidence
30	Muse Spark 1.2	44.6%	estimated ± 5.1 pp, medium confidence
31	Ornith-1.5-397B	44.6%	measured
32	Grok 4.5	44.5%	estimated ± 5.0 pp, medium confidence
33	Qwen3.8 Max	43.6%	measured
34	Kimi K3	43.5%	measured
35	GPT-5.6 Terra	43.5%	estimated ± 4.4 pp, high confidence
36	Hy4 preview	43.4%	measured
37	Claude Sonnet 5	43.2%	measured
38	GPT-5.5 Pro	43.1%	measured
39	dots3-note Preview	42.9%	estimated ± 5.3 pp, medium confidence
40	Muse Spark	42.8%	measured
41	GPT-5.4 Pro	42.7%	measured
42	Pareto 26.10 Preview	42.4%	estimated ± 4.4 pp, high confidence
43	Grok 4.7	42.2%	estimated ± 5.7 pp, medium confidence
44	Qwen3.8 Max Preview	42.2%	estimated ± 5.7 pp, medium confidence
45	GPT-5.6 Luna	42.1%	estimated ± 4.4 pp, high confidence
46	GPT-5.2	41.9%	estimated ± 4.9 pp, high confidence
47	GPT-5.3 Codex	41.4%	estimated ± 5.7 pp, medium confidence
48	GPT-5.5	41.4%	measured
49	GLM-5.2	40.5%	measured
50	Claude Opus 4.6	40.0%	measured
51	GPT-5.4	39.8%	measured
52	Qwen3.7 Max	39.6%	estimated ± 2.9 pp, high confidence
53	Gemini 3.5 Flash	39.0%	estimated ± 2.9 pp, high confidence
54	Agents-A1	38.8%	estimated ± 5.3 pp, medium confidence
55	Claude Opus 4.6 (Adaptive)	38.3%	estimated ± 5.7 pp, medium confidence
56	Gemini 3 Pro	38.0%	estimated ± 5.7 pp, medium confidence
57	Ling 3.1 Flash	37.7%	estimated ± 5.7 pp, medium confidence
58	GLM-5.3	37.0%	estimated ± 5.1 pp, medium confidence
59	Gemini 3 Pro Deep Think	36.3%	estimated ± 6.9 pp, medium confidence
60	Beam	36.2%	measured
61	Qwen3.8-Flash-Next	35.9%	measured
62	Interfaze Beta	35.6%	estimated ± 4.4 pp, high confidence
63	Qwen3.6 Plus	35.4%	estimated ± 4.9 pp, high confidence
64	Qwen3.8-Omni-Flash	35.2%	estimated ± 2.9 pp, high confidence
65	DeepSeek V4.1 Flash	35.1%	estimated ± 2.9 pp, high confidence
66	DeepSeek V4 Pro 0813	34.6%	estimated ± 2.9 pp, high confidence
67	Grok 4.3	34.5%	estimated ± 4.9 pp, high confidence
68	Gemini 3.1 Flash-Lite	34.2%	estimated ± 5.1 pp, medium confidence
69	MiMo-V2.5-Pro	34.0%	measured
70	Claude Sonnet 4.6	33.9%	estimated ± 4.9 pp, high confidence
71	GLM-5.3-Flash	33.8%	estimated ± 5.1 pp, medium confidence
72	GPT-6 Luna	33.8%	estimated ± 5.0 pp, medium confidence
73	Kimi K2.6	33.7%	estimated ± 2.9 pp, high confidence
74	GPT-5.2-Codex	33.4%	estimated ± 5.7 pp, medium confidence
75	Qwen3.7 Plus	33.2%	estimated ± 2.9 pp, high confidence
76	Solar Pro 4	32.8%	estimated ± 4.4 pp, high confidence
77	MiMo-V2.6-Flash	32.7%	estimated ± 5.7 pp, medium confidence
78	Gemini 3.5 Flash-Lite	32.6%	estimated ± 5.1 pp, medium confidence
79	Kimi K2.7 Code	32.6%	estimated ± 5.7 pp, medium confidence
80	Mistral Large 4	32.6%	estimated ± 5.7 pp, medium confidence
81	Grok 4.20	31.6%	measured
82	Inkling-Small	31.6%	measured
83	Apodex 1.1 Mini	31.6%	estimated ± 5.7 pp, medium confidence
84	Hy3	30.9%	estimated ± 5.7 pp, medium confidence
85	Qwen3.8-27B	30.8%	measured
86	Inkling	30.0%	measured
87	Qwen3.5 397B	29.3%	estimated ± 4.9 pp, high confidence
88	Solar Open 2	28.8%	measured
89	MiniMax M3	28.3%	estimated ± 5.1 pp, medium confidence
90	Claude Sonnet 4.5 Thinking	28.2%	estimated ± 6.0 pp, medium confidence
91	GPT-5.4 mini	28.2%	measured
92	DeepSeek V4 Flash 0731	28.2%	estimated ± 2.9 pp, high confidence
93	Qwen3.5 Flash	28.1%	estimated ± 5.1 pp, medium confidence
94	Qwen 3.6 Max (preview)	27.9%	estimated ± 5.7 pp, medium confidence
95	Qwen3.6-27B	27.6%	estimated ± 4.9 pp, high confidence
96	MiMo-V2-Pro	27.5%	estimated ± 5.7 pp, medium confidence
97	o3-pro	27.5%	estimated ± 6.3 pp, medium confidence
98	Claude Opus 4.5 Thinking	27.2%	estimated ± 5.7 pp, medium confidence
99	Kimi K2.5 (Reasoning)	27.0%	estimated ± 4.9 pp, high confidence
100	GLM-5	27.0%	estimated ± 2.9 pp, high confidence
101	Nemotron 3 Ultra	26.7%	measured
102	MiMo-V2.5	26.5%	estimated ± 5.1 pp, medium confidence
103	MiniMax M2.7	26.3%	estimated ± 4.4 pp, high confidence
104	GLM-4.6	26.0%	estimated ± 5.1 pp, medium confidence
105	Kimi K2.5	25.9%	estimated ± 2.9 pp, high confidence
106	Ornith-1.5-35B-A3B	25.6%	measured
107	GLM-4.5	25.6%	estimated ± 5.1 pp, medium confidence
108	GPT-5.1	25.5%	estimated ± 5.7 pp, medium confidence
109	GPT-5 (high)	25.5%	estimated ± 5.7 pp, medium confidence
110	Claude Opus 4.5	25.4%	estimated ± 4.9 pp, high confidence
111	Claude Haiku 4.5	25.2%	estimated ± 5.1 pp, medium confidence
112	Mistral Medium 3.5 128B	25.1%	estimated ± 5.1 pp, low confidence
113	Laguna M.1	25.1%	estimated ± 5.1 pp, low confidence
114	Laguna XS.2	25.1%	estimated ± 5.1 pp, low confidence
115	GLM-5-Turbo	24.8%	estimated ± 5.7 pp, medium confidence
116	Qwen3.5-122B-A10B	24.3%	estimated ± 4.9 pp, high confidence
117	GPT-5.4 nano	24.3%	measured
118	Hy3 Preview	24.0%	estimated ± 2.9 pp, high confidence
119	Claude 4.1 Opus	23.8%	estimated ± 6.3 pp, medium confidence
120	GLM-5.1	23.8%	estimated ± 4.4 pp, high confidence
121	Grok 4	23.6%	estimated ± 5.7 pp, medium confidence
122	Qwen3.6-35B-A3B	22.9%	estimated ± 4.9 pp, high confidence
123	GPT-5.1-Codex	22.6%	estimated ± 5.7 pp, medium confidence
124	GPT-5.1-Codex-Max	22.6%	estimated ± 5.7 pp, medium confidence
125	Ternary Bonsai 2 27B	22.5%	estimated ± 4.4 pp, high confidence
126	GPT-5 (medium)	22.3%	estimated ± 5.7 pp, medium confidence
127	GLM-4.7	22.1%	estimated ± 4.9 pp, high confidence
128	A.X K2	22.0%	estimated ± 4.4 pp, high confidence
129	Qwen3.5-27B	21.7%	estimated ± 4.9 pp, high confidence
130	Ornith-1.5-9B	20.2%	measured
131	Gemma 4 31B	19.5%	measured
132	MiMo-V2-Omni	19.1%	estimated ± 5.7 pp, medium confidence
133	Ling 3.0 Flash VL	19.0%	estimated ± 5.7 pp, medium confidence
134	Muse Glimmer 30B	19.0%	estimated ± 5.7 pp, medium confidence
135	Qwen3.5-35B-A3B	18.9%	estimated ± 4.9 pp, high confidence
136	Step 3.7 Flash	18.4%	estimated ± 5.7 pp, medium confidence
137	Ling 3.0 Flash	18.2%	estimated ± 2.9 pp, high confidence
138	MAI-Thinking-1	18.2%	estimated ± 4.4 pp, high confidence
139	MiMo-V2-Flash	17.9%	estimated ± 4.9 pp, high confidence
140	Nemotron 3 Super 100B	17.8%	estimated ± 5.7 pp, medium confidence
141	Ling 3.0 Flash FP8	17.7%	estimated ± 4.4 pp, high confidence
142	Claude Sonnet 4.5	17.4%	estimated ± 4.9 pp, high confidence
143	o3	17.2%	estimated ± 5.7 pp, medium confidence
144	Qwen3.5 397B (Reasoning)	16.9%	estimated ± 5.7 pp, medium confidence
145	GPT-OSS 120B	16.7%	estimated ± 5.7 pp, medium confidence
146	Gemini 2.5 Pro	16.7%	estimated ± 4.9 pp, high confidence
147	Grok 4.1 Fast (Reasoning)	16.4%	estimated ± 5.7 pp, medium confidence
148	Grok 4 Fast (Reasoning)	16.2%	estimated ± 5.7 pp, medium confidence
149	Quasar 438B	15.8%	estimated ± 5.7 pp, medium confidence
150	o1-preview	15.3%	estimated ± 6.3 pp, low confidence
151	GLM-5V-Turbo	14.4%	estimated ± 5.7 pp, medium confidence
152	K-EXAONE 2.0	13.8%	estimated ± 4.4 pp, high confidence
153	DeepSeek-R1	13.2%	estimated ± 5.7 pp, medium confidence
154	DeepSeek V3.1 (Reasoning)	11.8%	estimated ± 5.7 pp, medium confidence
155	K-Exaone	11.5%	estimated ± 5.7 pp, medium confidence
156	o1-pro	11.4%	estimated ± 4.9 pp, high confidence
157	Nemotron 3.5 Lightning 30B A3B NVFP4	10.5%	measured
158	Claude 4.1 Opus Thinking	10.2%	estimated ± 5.7 pp, medium confidence
159	Qwen3 235B 2507	10.2%	estimated ± 4.9 pp, high confidence
160	o3-mini	9.9%	estimated ± 4.9 pp, high confidence
161	Command A+	9.8%	estimated ± 5.7 pp, medium confidence
162	Qwen3 Max	9.7%	estimated ± 5.7 pp, medium confidence
163	Mercury 2.5	9.5%	estimated ± 4.4 pp, high confidence
164	Nemotron 3 Nano 30B	9.3%	estimated ± 5.7 pp, medium confidence
165	DeepSeek V3.2	9.1%	estimated ± 5.7 pp, medium confidence
166	North Mini Code	9.0%	estimated ± 5.7 pp, medium confidence
167	o1	9.0%	estimated ± 4.9 pp, high confidence
168	GPT-OSS 20B	8.9%	estimated ± 5.7 pp, medium confidence
169	Sarvam 105B	8.9%	estimated ± 5.7 pp, medium confidence
170	Gemma 4 26B A4B	8.7%	measured
171	Solar Pro 3	8.3%	estimated ± 5.7 pp, low confidence
172	Mistral Small 4	8.0%	estimated ± 5.7 pp, low confidence
173	Mistral Small 4 (Reasoning)	8.0%	estimated ± 5.7 pp, low confidence
174	Trinity-Large-Thinking	7.6%	estimated ± 4.4 pp, high confidence
175	Ling 3.0 Tiny	7.5%	estimated ± 5.7 pp, low confidence
176	Granite 4.2 30B	6.6%	estimated ± 4.9 pp, medium confidence
177	GPT-4.1	6.5%	estimated ± 4.9 pp, medium confidence
178	Grok Code Fast 1	6.4%	estimated ± 5.7 pp, low confidence
179	GPT-4.1 mini	6.4%	estimated ± 4.9 pp, medium confidence
180	Granite 4.2 8B	6.4%	estimated ± 4.9 pp, medium confidence
181	Nemotron 3 Nano Omni 30B A3B	6.4%	estimated ± 4.4 pp, medium confidence
182	ZAYA1-8B	6.2%	estimated ± 4.4 pp, medium confidence
183	Claude 3.5 Sonnet	6.2%	estimated ± 4.9 pp, medium confidence
184	DeepSeek V3	6.2%	estimated ± 4.9 pp, medium confidence
185	Ling 2.6 Flash	6.2%	estimated ± 4.9 pp, medium confidence
186	Gemma 4 E4B	6.1%	estimated ± 4.9 pp, medium confidence
187	MiniCPM5-2B	6.1%	estimated ± 4.4 pp, medium confidence
188	Granite 4.2 3B	6.1%	estimated ± 4.9 pp, medium confidence
189	GPT-4.1 nano	6.0%	estimated ± 4.9 pp, medium confidence
190	LongCat-Flash-Lite-Sparse	6.0%	estimated ± 4.4 pp, medium confidence
191	Gemma 4 E2B	6.0%	estimated ± 4.9 pp, medium confidence
192	LFM2.5-VL-450M	6.0%	estimated ± 4.9 pp, medium confidence
193	Qwen3-Omni-30B-A3B-Thinking	6.0%	estimated ± 5.7 pp, low confidence
194	Sarvam 30B	6.0%	estimated ± 5.7 pp, low confidence
195	Kimi K2	5.9%	estimated ± 5.7 pp, low confidence
196	Nemotron Ultra 253B	5.9%	estimated ± 5.7 pp, low confidence
197	Trinity-Large-Preview	5.8%	estimated ± 4.4 pp, medium confidence
198	Mellum2-12B-A2.5B-Thinking	5.8%	estimated ± 4.4 pp, medium confidence
199	ZAYA1-74B-Preview	5.7%	estimated ± 4.4 pp, medium confidence
200	LLaDA2.2-mini	5.7%	estimated ± 4.4 pp, medium confidence
201	Soofi S 30B-A3B	5.7%	estimated ± 4.4 pp, medium confidence
202	Mellum2-12B-A2.5B-Instruct	5.7%	estimated ± 4.4 pp, medium confidence
203	LFM2.5-230M	5.7%	estimated ± 4.4 pp, medium confidence
204	MiniCPM5-1B	5.7%	estimated ± 4.4 pp, medium confidence
205	GLM-4.5-Air	5.6%	estimated ± 5.7 pp, low confidence
206	LFM2.5-8B-A1B	5.5%	estimated ± 5.7 pp, low confidence
207	Celeris-1	5.4%	estimated ± 5.7 pp, low confidence
208	DeepSeek V3.1	5.3%	estimated ± 5.7 pp, low confidence
209	Gemma 4 12B	5.2%	measured
210	Granite-4.0-H-350M	5.1%	estimated ± 5.7 pp, low confidence
211	LFM2.5-2.6B	4.9%	estimated ± 5.7 pp, low confidence
212	Exaone 4.0 1.2B	4.5%	estimated ± 5.7 pp, low confidence
213	Granite-4.0-350M	4.3%	estimated ± 5.7 pp, low confidence
214	Grok 4.1 Fast	4.0%	estimated ± 5.7 pp, low confidence
215	LFM2.5-VL-1.6B-Extract	4.0%	estimated ± 5.7 pp, low confidence
216	Exaone 4.0 32B	3.9%	estimated ± 5.7 pp, low confidence
217	Granite-4.0-H-1B	3.9%	estimated ± 5.7 pp, low confidence
218	Phi-4 Multimodal Instruct	3.9%	estimated ± 5.7 pp, low confidence
219	Llama 4 Maverick	3.9%	estimated ± 5.7 pp, low confidence
220	DeepSeek V3 0324	3.7%	estimated ± 5.7 pp, low confidence
221	Gemini 2.5 Flash	3.7%	estimated ± 5.7 pp, low confidence
222	DeepSeek R1 Distill Qwen 32B	3.6%	estimated ± 5.7 pp, low confidence
223	Gemini 1.5 Pro	3.6%	estimated ± 5.7 pp, low confidence
224	Qwen3-Omni-30B-A3B-Instruct	3.6%	estimated ± 5.7 pp, low confidence
225	Gemma 3 27B	3.5%	estimated ± 5.7 pp, low confidence
226	Claude 4 Sonnet	3.4%	estimated ± 5.7 pp, low confidence
227	Gemini 1.0 Pro	3.3%	estimated ± 5.7 pp, low confidence
228	GPT-4o mini	3.3%	estimated ± 5.7 pp, low confidence
229	Mistral Large 3	3.3%	estimated ± 5.7 pp, low confidence
230	Claude 3 Haiku	3.2%	estimated ± 5.7 pp, low confidence
231	Mistral Medium 3	3.2%	estimated ± 5.7 pp, low confidence
232	Llama 3.1 405B	3.1%	estimated ± 5.7 pp, low confidence
233	Llama 4 Scout	3.0%	estimated ± 5.7 pp, low confidence
234	Phi-4	3.0%	estimated ± 5.7 pp, low confidence
235	Solar Pro 2	2.9%	estimated ± 5.7 pp, low confidence
236	Ultravox v0.6 Llama 3.3 70B	2.8%	estimated ± 5.7 pp, low confidence
237	Qwen2.5 Coder 32B Instruct	2.7%	estimated ± 5.7 pp, low confidence
238	Mistral Large 2	2.6%	estimated ± 5.7 pp, low confidence
239	Nova Pro	2.5%	estimated ± 5.7 pp, low confidence
240	GPT-4 Turbo	2.4%	estimated ± 5.7 pp, low confidence
241	Claude 3 Opus	2.2%	estimated ± 5.7 pp, low confidence
242	GPT-4o	1.9%	estimated ± 5.7 pp, low confidence

HLE w/o tools leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General