Agentic · tools

HLE w/ tools leaderboard

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

As of 2026-10-07, the highest measured score on HLE w/ tools is 67.7% by Claude Opus 5.5. 131 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Muse Spark 1.3	73.5%	estimated ± 4.6 pp, low confidence
2	MiMo-V2.6-Flash	72.7%	estimated ± 5.0 pp, low confidence
3	MiMo-V2.6-Pro	71.7%	estimated ± 5.0 pp, low confidence
4	Claude Opus 5.5	67.7%	measured
5	Claude Fable 5	65.8%	estimated ± 5.7 pp, low confidence
6	Ling 3.1 Flash	65.4%	estimated ± 5.0 pp, low confidence
7	Grok 4.7	65.2%	estimated ± 4.6 pp, medium confidence
8	Claude Opus 5	64.7%	measured
9	GLM-5.2	64.6%	estimated ± 4.6 pp, medium confidence
10	Claude Sonnet 5.5	64.5%	measured
11	Fugu Cyber	64.2%	estimated ± 5.0 pp, low confidence
12	DeepSeek V4.1 Flash	63.9%	measured
13	Gemini 3.8 Flash Cyber	63.3%	estimated ± 5.0 pp, low confidence
14	GLM-5.3	62.5%	measured
15	Qwen3.8 Max Preview	61.7%	estimated ± 6.1 pp, medium confidence
16	Gemini 4 Argon	60.5%	estimated ± 6.1 pp, medium confidence
17	Grok 4.6	60.4%	estimated ± 6.1 pp, medium confidence
18	Gemini 3.7 Flash	60.2%	estimated ± 5.7 pp, low confidence
19	DeepSeek V4 Pro 0813	60.0%	measured
20	GPT-5.6 Sol	59.3%	estimated ± 4.6 pp, medium confidence
21	Gemini 3.5 Flash Cyber	59.3%	estimated ± 5.0 pp, low confidence
22	Kimi K3	59.3%	estimated ± 4.6 pp, high confidence
23	GPT-6.1 Sol	59.2%	estimated ± 6.1 pp, medium confidence
24	GPT-5.5 Pro	59.1%	estimated ± 4.6 pp, high confidence
25	Claude Mythos Preview	59.1%	estimated ± 5.0 pp, low confidence
26	GPT-5.4 Pro	59.0%	estimated ± 4.6 pp, high confidence
27	Step 5 Preview	58.9%	estimated ± 4.6 pp, high confidence
28	Claude Mythos 5	58.8%	estimated ± 4.6 pp, high confidence
29	GPT-5.6 Terra	58.7%	estimated ± 4.6 pp, high confidence
30	GPT-6 Sol	58.3%	estimated ± 4.6 pp, medium confidence
31	Claude Fable 5.1	58.2%	estimated ± 4.6 pp, medium confidence
32	Qwen3.8-Flash-Next	58.0%	estimated ± 5.9 pp, medium confidence
33	GPT-5.5	57.6%	estimated ± 4.6 pp, high confidence
34	Claude Opus 4.8	57.5%	estimated ± 4.6 pp, high confidence
35	Claude Haiku 5.5	57.4%	measured
36	Claude Sonnet 5	57.4%	measured
37	Claude Opus 4.6	57.2%	estimated ± 4.6 pp, high confidence
38	GPT-6 Astra	57.2%	measured
39	MiniMax M3	57.1%	estimated ± 4.6 pp, high confidence
40	GPT-5.6 Luna	57.0%	estimated ± 4.6 pp, high confidence
41	Muse Spark 1.2	56.7%	estimated ± 6.1 pp, medium confidence
42	GPT-5.4	56.6%	estimated ± 4.6 pp, high confidence
43	Qwen3.8 Max	56.2%	measured
44	Apodex 1.1	56.1%	measured
45	Ornith-1.5-397B	56.1%	measured
46	GPT-6 Luna	55.7%	estimated ± 6.1 pp, medium confidence
47	Atria Dawn Preview	55.6%	estimated ± 2.4 pp, medium confidence
48	Gemini 3.8 Flash	55.4%	estimated ± 4.6 pp, low confidence
49	Hy4 preview	55.4%	measured
50	Mistral Large 4	55.3%	estimated ± 6.1 pp, medium confidence
51	Qwen3.8-27B	55.3%	estimated ± 6.1 pp, medium confidence
52	GLM-5.3-Flash	55.3%	measured
53	Kimi K2.6	55.0%	estimated ± 2.4 pp, medium confidence
54	Grok 4.5	54.4%	estimated ± 6.1 pp, medium confidence
55	Qwen3.7 Max	53.5%	measured
56	Gemini 3.5 Flash	53.2%	estimated ± 6.1 pp, medium confidence
57	Claude Opus 4.7 (Adaptive)	53.1%	estimated ± 4.6 pp, high confidence
58	dots3-note Preview	52.6%	measured
59	Gemini 3.6 Flash	51.8%	estimated ± 6.1 pp, medium confidence
60	Inkling-Small	50.3%	estimated ± 4.6 pp, high confidence
61	Beam	50.3%	estimated ± 4.6 pp, high confidence
62	Hy3 Preview	50.0%	estimated ± 6.1 pp, medium confidence
63	Inkling	49.8%	estimated ± 4.6 pp, high confidence
64	Claude Opus 4.5	49.6%	estimated ± 2.4 pp, medium confidence
65	Apodex 1.1 Mini	49.4%	estimated ± 6.1 pp, medium confidence
66	Quasar 438B	49.1%	estimated ± 6.1 pp, medium confidence
67	Ling 3.0 Flash VL	48.6%	estimated ± 6.1 pp, medium confidence
68	MiMo-V2.5-Pro	47.6%	estimated ± 6.1 pp, medium confidence
69	Agents-A1	47.6%	measured
70	Step 3.7 Flash	47.2%	measured
71	Grok 4.3	46.6%	estimated ± 6.1 pp, medium confidence
72	Hy3	46.1%	estimated ± 6.1 pp, medium confidence
73	Kimi K2.7 Code	45.4%	estimated ± 6.1 pp, medium confidence
74	DeepSeek V4 Flash 0731	45.1%	measured
75	Qwen3.6 Plus	45.0%	estimated ± 2.4 pp, medium confidence
76	GPT-5.4 mini	44.8%	estimated ± 6.1 pp, medium confidence
77	MiniMax M2.7	44.8%	estimated ± 6.1 pp, low confidence
78	Qwen3.5 397B	44.3%	estimated ± 2.4 pp, medium confidence
79	Qwen3.6-27B	44.1%	estimated ± 6.1 pp, low confidence
80	Gemini 3.5 Flash-Lite	44.1%	estimated ± 6.1 pp, low confidence
81	A.X K2	43.5%	estimated ± 6.1 pp, low confidence
82	Ling 3.0 Flash	43.4%	estimated ± 2.4 pp, medium confidence
83	GPT-5.4 nano	43.1%	estimated ± 6.1 pp, low confidence
84	Ling 3.0 Flash FP8	43.1%	estimated ± 6.1 pp, low confidence
85	GPT-5 (high)	42.4%	estimated ± 6.1 pp, low confidence
86	Kimi K2.5	41.3%	estimated ± 2.4 pp, medium confidence
87	GPT-5.1	40.1%	estimated ± 6.1 pp, low confidence
88	Gemini 3.1 Pro	39.1%	estimated ± 6.1 pp, low confidence
89	Muse Glimmer 30B	39.0%	estimated ± 6.1 pp, low confidence
90	Qwen3.7 Plus	38.5%	estimated ± 6.1 pp, low confidence
91	Mistral Medium 3.5 128B	38.4%	estimated ± 6.1 pp, low confidence
92	Laguna S 2.1	38.3%	estimated ± 5.9 pp, medium confidence
93	Nemotron 3 Ultra	37.4%	measured
94	MiniCPM5-2B	37.0%	estimated ± 6.1 pp, low confidence
95	GLM-5.1	36.6%	estimated ± 4.6 pp, high confidence
96	Agents-A1-4B	35.7%	estimated ± 4.6 pp, high confidence
97	GLM-5	35.7%	estimated ± 2.4 pp, medium confidence
98	GPT-5.2	35.2%	estimated ± 4.6 pp, high confidence
99	MiMo-V2-Flash	35.0%	estimated ± 6.1 pp, low confidence
100	Gemma 4 31B	34.7%	estimated ± 6.1 pp, low confidence
101	GPT-OSS 120B	34.5%	estimated ± 6.1 pp, low confidence
102	Qwen3.5-122B-A10B	34.4%	estimated ± 4.6 pp, high confidence
103	Qwen3.5-27B	33.8%	estimated ± 4.6 pp, high confidence
104	Qwen3.5-35B-A3B	33.8%	estimated ± 4.6 pp, high confidence
105	Kimi K2.5 (Reasoning)	33.7%	estimated ± 4.6 pp, high confidence
106	Granite 4.2 30B	33.6%	estimated ± 6.1 pp, low confidence
107	Solar Open 2	33.5%	estimated ± 7.2 pp, medium confidence
108	Ornith-1.5-35B-A3B	33.4%	measured
109	Gemma 4 26B A4B	33.3%	estimated ± 6.1 pp, low confidence
110	GLM-4.7	33.2%	estimated ± 4.6 pp, high confidence
111	Ling 3.0 Tiny	33.2%	estimated ± 6.1 pp, low confidence
112	Solar Pro 4	33.2%	estimated ± 4.6 pp, high confidence
113	LongCat-Flash-Lite-Sparse	33.2%	estimated ± 4.6 pp, high confidence
114	Nemotron 3.5 Lightning 30B A3B NVFP4	33.2%	estimated ± 4.6 pp, medium confidence
115	Command A+	32.0%	estimated ± 6.1 pp, low confidence
116	Celeris-1	31.6%	estimated ± 6.1 pp, low confidence
117	DeepSeek V3	31.6%	estimated ± 6.1 pp, low confidence
118	DeepSeek V3 0324	31.6%	estimated ± 6.1 pp, low confidence
119	Gemini 2.5 Pro	31.6%	estimated ± 6.1 pp, low confidence
120	Gemma 3 27B	31.6%	estimated ± 6.1 pp, low confidence
121	Gemma 4 12B	31.6%	estimated ± 6.1 pp, low confidence
122	Gemma 4 E2B	31.6%	estimated ± 6.1 pp, low confidence
123	Gemma 4 E4B	31.6%	estimated ± 6.1 pp, low confidence
124	GPT-4.1 mini	31.6%	estimated ± 6.1 pp, low confidence
125	GPT-4.1 nano	31.6%	estimated ± 6.1 pp, low confidence
126	GPT-4o	31.6%	estimated ± 6.1 pp, low confidence
127	GPT-4o mini	31.6%	estimated ± 6.1 pp, low confidence
128	GPT-OSS 20B	31.6%	estimated ± 6.1 pp, low confidence
129	Granite 4.2 3B	31.6%	estimated ± 6.1 pp, low confidence
130	Granite 4.2 8B	31.6%	estimated ± 6.1 pp, low confidence
131	K-Exaone	31.6%	estimated ± 6.1 pp, low confidence
132	LFM2.5-2.6B	31.6%	estimated ± 6.1 pp, low confidence
133	Ling 2.6 Flash	31.6%	estimated ± 6.1 pp, low confidence
134	Llama 4 Maverick	31.6%	estimated ± 6.1 pp, low confidence
135	Llama 4 Scout	31.6%	estimated ± 6.1 pp, low confidence
136	Mercury 2.5	31.6%	estimated ± 6.1 pp, low confidence
137	Mistral Large 3	31.6%	estimated ± 6.1 pp, low confidence
138	Mistral Small 4	31.6%	estimated ± 6.1 pp, low confidence
139	Mistral Small 4 (Reasoning)	31.6%	estimated ± 6.1 pp, low confidence
140	Nemotron 3 Nano 30B	31.6%	estimated ± 6.1 pp, low confidence
141	Nemotron 3 Nano Omni 30B A3B	31.6%	estimated ± 6.1 pp, low confidence
142	Nemotron 3 Super 100B	31.6%	estimated ± 6.1 pp, low confidence
143	North Mini Code	31.6%	estimated ± 6.1 pp, low confidence
144	Solar Pro 3	31.6%	estimated ± 6.1 pp, low confidence
145	Trinity-Large-Preview	31.6%	estimated ± 6.1 pp, low confidence
146	Trinity-Large-Thinking	31.6%	estimated ± 6.1 pp, low confidence
147	Ultravox v0.6 Llama 3.3 70B	31.6%	estimated ± 6.1 pp, low confidence
148	Qwen3.6-35B-A3B	30.6%	estimated ± 2.4 pp, medium confidence
149	Ornith-1.5-9B	30.5%	measured
150	Claude Sonnet 4.6	28.1%	estimated ± 5.0 pp, low confidence
151	LLaDA2.2-flash	23.9%	estimated ± 7.2 pp, low confidence
152	Muse Spark 1.1	18.0%	estimated ± 5.0 pp, low confidence
153	Muse Spark	3.5%	estimated ± 5.0 pp, low confidence

HLE w/ tools leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General