benchgap
Agentic · tools

HLE w/ tools leaderboard

As of 2026-11-21, the highest measured score on HLE w/ tools is 70.0% by Kimi K3.1 Pro. 176 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Kimi K3.1 Pro70.0%measured
2Claude Opus 5.567.7%measured
3GPT-6.1 Terra67.0%estimated ± 6.8 pp, low confidence
4Gemini 4 Carbon66.9%estimated ± 6.8 pp, low confidence
5Gemini 4 Neon66.7%estimated ± 6.8 pp, low confidence
6Gemini 4 Hydrogen66.4%estimated ± 6.8 pp, low confidence
7Qwen4 Max65.5%estimated ± 6.8 pp, low confidence
8Claude Fable 5.165.3%estimated ± 4.9 pp, medium confidence
9Claude Opus 564.7%measured
10GPT-6.2 Sol64.6%estimated ± 5.1 pp, low confidence
11GPT-6.2 Astra64.6%estimated ± 5.1 pp, low confidence
12Claude Sonnet 5.564.5%measured
13Claude Fable 5.264.2%estimated ± 6.8 pp, low confidence
14DeepSeek V4.1 Flash 101763.9%measured
15MiMo-V2.6-Flash63.8%estimated ± 4.9 pp, low confidence
16MiMo-V2.6-Pro63.8%estimated ± 4.9 pp, low confidence
17Muse Wonder Full63.7%estimated ± 4.9 pp, medium confidence
18Ling 3.1 Flash63.6%estimated ± 4.9 pp, medium confidence
19Fugu Cyber63.4%estimated ± 4.9 pp, medium confidence
20Atria Dawn Preview63.3%estimated ± 4.9 pp, medium confidence
21Gemini 3.8 Flash Cyber63.2%estimated ± 4.9 pp, medium confidence
22Grok 4.763.1%estimated ± 3.1 pp, high confidence
23LongCat-2.563.1%measured
24GLM-5.362.5%measured
25GPT-6.1 Sol62.4%estimated ± 3.1 pp, high confidence
26Inkling 2 Small62.4%estimated ± 5.6 pp, medium confidence
27Muse Spark 1.362.3%estimated ± 5.3 pp, medium confidence
28GPT-5.5 Pro62.3%estimated ± 5.6 pp, medium confidence
29GPT-5.6 Sol62.2%estimated ± 4.9 pp, medium confidence
30GPT-6 Sol61.7%estimated ± 3.1 pp, high confidence
31GPT-5.4 Pro61.5%estimated ± 5.6 pp, medium confidence
32Claude Mythos 561.5%estimated ± 4.9 pp, medium confidence
33Qwen3.8-Flash-Next61.1%estimated ± 5.3 pp, medium confidence
34Muse Spark Open61.0%measured
35Gemini 4 Argon61.0%estimated ± 5.3 pp, medium confidence
36Grok 4.660.8%estimated ± 5.3 pp, medium confidence
37Gemini 3.5 Flash Cyber60.7%estimated ± 4.9 pp, medium confidence
38Claude Fable 560.6%estimated ± 5.3 pp, medium confidence
39Claude Mythos Preview60.6%estimated ± 4.9 pp, medium confidence
40Muse Wonder60.3%estimated ± 4.9 pp, medium confidence
41DeepSeek V4 Pro60.0%measured
42ERNIE 5.5 Flash Preview59.8%estimated ± 5.6 pp, medium confidence
43Hy459.1%measured
44MiniMax M3.1 Flash59.0%estimated ± 4.9 pp, medium confidence
45Muse Ember58.8%estimated ± 4.9 pp, medium confidence
46Kimi K358.5%estimated ± 3.1 pp, high confidence
47Claude Haiku 5.557.4%measured
48Claude Sonnet 557.4%measured
49Muse Spark 1.257.3%estimated ± 5.3 pp, medium confidence
50GPT-6 Astra57.2%measured
51Ling 3.0 Flash Sante57.1%measured
52Gemini 3.5 Flash57.0%estimated ± 4.1 pp, low confidence
53Qwen3.8 Max Preview57.0%estimated ± 4.1 pp, low confidence
54GPT-5.6 Terra57.0%estimated ± 4.1 pp, low confidence
55GPT-5.557.0%estimated ± 4.1 pp, low confidence
56GPT-5.6 Luna57.0%estimated ± 4.1 pp, low confidence
57GPT-5.457.0%estimated ± 4.1 pp, medium confidence
58Claude Opus 4.6 (Adaptive)57.0%estimated ± 4.1 pp, medium confidence
59Gemini 3.1 Pro56.9%estimated ± 4.1 pp, medium confidence
60Apodex 1.1 Mini56.9%estimated ± 4.1 pp, medium confidence
61GPT-6 Luna56.9%estimated ± 3.1 pp, high confidence
62Claude Opus 4.856.9%estimated ± 5.3 pp, medium confidence
63Kimi K2.656.8%estimated ± 4.1 pp, medium confidence
64GPT-5.4 mini56.8%estimated ± 4.1 pp, medium confidence
65Ling 3.0 Flash Fin56.8%estimated ± 4.1 pp, medium confidence
66GPT-5.4 nano56.6%estimated ± 4.1 pp, medium confidence
67Qwen3.7 Plus56.2%estimated ± 4.1 pp, medium confidence
68Qwen3.8 Max56.2%measured
69Apodex 1.156.1%measured
70Ornith-1.5-397B56.1%measured
71Hy4 preview55.4%measured
72Gemini 3.7 Flash55.4%estimated ± 5.3 pp, medium confidence
73GLM-5.3-Flash55.3%measured
74Hy4 flash55.2%measured
75Grok 4.555.2%estimated ± 5.3 pp, medium confidence
76Sage-A1.555.1%measured
77Step 5 Preview55.0%estimated ± 3.1 pp, high confidence
78GLM-5.254.7%measured
79Claude Haiku 4.553.7%estimated ± 7.5 pp, low confidence
80GPT-5.3-Codex-Spark53.7%estimated ± 7.5 pp, low confidence
81Grok 4.2053.7%estimated ± 7.5 pp, low confidence
82GLM-5-Turbo53.5%estimated ± 5.5 pp, medium confidence
83GLM-5V-Turbo53.5%estimated ± 5.5 pp, medium confidence
84Qwen3.7 Max53.5%measured
85Grok 4.353.1%estimated ± 4.1 pp, medium confidence
86Gemini 3.6 Flash52.7%estimated ± 5.3 pp, medium confidence
87dots3-note Preview52.6%measured
88Inkling 2 Mini52.3%estimated ± 5.6 pp, medium confidence
89GLM-5.152.3%measured
90Ling 3.0 Flash52.3%estimated ± 4.3 pp, low confidence
91MiniMax M351.7%estimated ± 5.3 pp, medium confidence
92Mistral Large 451.3%estimated ± 5.1 pp, medium confidence
93Qwen3.8-27B51.2%estimated ± 5.1 pp, medium confidence
94Qwen 3.6 Max (preview)51.1%estimated ± 5.5 pp, medium confidence
95Beam 1.0 501B-A23B51.1%measured
96Beam 1.0 501B-A23B FP851.1%measured
97Beam 1.0 501B-A23B NVFP451.1%measured
98Hy3 Preview51.1%estimated ± 5.3 pp, medium confidence
99Qwen3.5 397B50.4%estimated ± 4.1 pp, medium confidence
100GLM-550.4%measured
101Quasar 438B50.3%estimated ± 5.3 pp, medium confidence
102MiMo-V2-Pro50.2%estimated ± 5.5 pp, medium confidence
103Claude Opus 4.750.1%estimated ± 4.9 pp, medium confidence
104Ripple-65B Preview50.0%estimated ± 9.7 pp, low confidence
105Claude Opus 4.650.0%estimated ± 4.9 pp, medium confidence
106Claude Sonnet 4.650.0%estimated ± 4.9 pp, medium confidence
107Muse Spark 1.150.0%estimated ± 4.9 pp, medium confidence
108Claude Opus 4.550.0%estimated ± 4.9 pp, medium confidence
109Muse Spark50.0%estimated ± 4.9 pp, medium confidence
110Nemotron 3.5 Super VL 120B A12B49.8%estimated ± 5.3 pp, medium confidence
111Nemotron 3.5 Super VL 120B A12B NVFP449.8%estimated ± 5.3 pp, medium confidence
112Ling 3.0 Flash VL49.7%estimated ± 5.3 pp, medium confidence
113Qwen3.5-27B49.1%estimated ± 5.5 pp, medium confidence
114Inkling-Small48.9%estimated ± 5.3 pp, medium confidence
115Grok 4.1 Fast (Reasoning)48.5%estimated ± 5.5 pp, medium confidence
116Solar Pro 448.4%estimated ± 5.3 pp, medium confidence
117Inkling47.6%estimated ± 5.3 pp, medium confidence
118Agents-A147.6%measured
119Hy347.4%estimated ± 5.3 pp, medium confidence
120Step 3.7 Flash47.2%measured
121GPT-5.2-Codex47.2%estimated ± 5.5 pp, medium confidence
122Qwen3.5-4B47.2%estimated ± 5.5 pp, medium confidence
123Kimi K2.7 Code46.8%estimated ± 5.3 pp, medium confidence
124GLM-4.746.2%estimated ± 5.3 pp, low confidence
125MiMo-V2-Omni46.2%estimated ± 5.5 pp, medium confidence
126Qwen3.6 Plus45.6%estimated ± 5.3 pp, low confidence
127Gemini 3.8 Flash45.5%estimated ± 3.1 pp, medium confidence
128Qwen3.6-27B45.5%estimated ± 5.3 pp, low confidence
129Gemini 3.5 Flash-Lite45.5%estimated ± 5.3 pp, low confidence
130Nemotron 3.5 Bolt 74B-A9B45.4%estimated ± 5.3 pp, low confidence
131Nemotron 3.5 Bolt 74B-A9B BF1645.3%estimated ± 5.6 pp, medium confidence
132DeepSeek V4 Flash 073145.1%measured
133A.X K244.9%estimated ± 5.3 pp, low confidence
134Claude Opus 4.5 Thinking44.2%estimated ± 5.5 pp, medium confidence
135GPT-543.9%estimated ± 5.3 pp, low confidence
136Qwen3.5-35B-A3B43.9%estimated ± 5.5 pp, medium confidence
137Qwen3.6-35B-A3B43.3%estimated ± 5.3 pp, low confidence
138Solar Open 243.2%estimated ± 6.3 pp, low confidence
139LFM2.5-8B-A1B42.5%estimated ± 5.5 pp, medium confidence
140GPT-5.141.7%estimated ± 5.3 pp, low confidence
141Qwen3.5-122B-A10B41.4%estimated ± 5.3 pp, low confidence
142Gemini 3 Pro41.3%estimated ± 5.5 pp, medium confidence
143K-EXAONE 2.041.2%estimated ± 5.3 pp, low confidence
144Muse Glimmer 30B40.6%estimated ± 5.3 pp, low confidence
145Mistral Medium 3.5 128B40.1%estimated ± 5.3 pp, low confidence
146GPT-5.3 Codex39.9%estimated ± 5.5 pp, medium confidence
147MiniCPM5-2B38.8%estimated ± 5.3 pp, low confidence
148Kimi K2.538.6%estimated ± 4.1 pp, low confidence
149Laguna S 2.138.4%estimated ± 5.8 pp, medium confidence
150GPT-5.238.4%estimated ± 5.5 pp, medium confidence
151Nemotron 3 Ultra37.4%measured
152Nemotron 3.5 Lightning 30B A3B NVFP437.1%estimated ± 5.3 pp, low confidence
153Spark-X2.6-4B36.9%measured
154MiMo-V2-Flash36.9%estimated ± 5.3 pp, low confidence
155Gemma 4 31B36.6%estimated ± 5.3 pp, low confidence
156Sarvam 105B36.4%estimated ± 5.6 pp, medium confidence
157LongCat-Flash-Lite-Sparse36.2%estimated ± 5.6 pp, medium confidence
158GPT-5.1-Codex36.0%estimated ± 5.5 pp, low confidence
159GPT-5.1-Codex-Max36.0%estimated ± 5.5 pp, low confidence
160Granite 4.2 30B35.6%estimated ± 5.3 pp, low confidence
161MiniMax M2.735.4%estimated ± 4.1 pp, low confidence
162Gemma 4 26B A4B35.3%estimated ± 5.3 pp, low confidence
163Ling 3.0 Tiny35.2%estimated ± 5.3 pp, low confidence
164Sarvam 30B34.7%estimated ± 5.6 pp, low confidence
165Command A+34.1%estimated ± 5.3 pp, low confidence
166Celeris-133.8%estimated ± 5.3 pp, low confidence
167Claude 3.5 Haiku33.8%estimated ± 5.3 pp, low confidence
168DeepSeek V333.8%estimated ± 5.3 pp, low confidence
169DeepSeek V3 032433.8%estimated ± 5.3 pp, low confidence
170Gemini 2.5 Pro33.8%estimated ± 5.3 pp, low confidence
171Gemma 3 27B33.8%estimated ± 5.3 pp, low confidence
172Gemma 4 12B33.8%estimated ± 5.3 pp, low confidence
173Gemma 4 E2B33.8%estimated ± 5.3 pp, low confidence
174Gemma 4 E4B33.8%estimated ± 5.3 pp, low confidence
175GPT-4.1 mini33.8%estimated ± 5.3 pp, low confidence
176GPT-4.1 nano33.8%estimated ± 5.3 pp, low confidence
177GPT-4o33.8%estimated ± 5.3 pp, low confidence
178GPT-4o mini33.8%estimated ± 5.3 pp, low confidence
179Granite 4.2 3B33.8%estimated ± 5.3 pp, low confidence
180Granite 4.2 8B33.8%estimated ± 5.3 pp, low confidence
181K-Exaone33.8%estimated ± 5.3 pp, low confidence
182LFM2.5-2.6B33.8%estimated ± 5.3 pp, low confidence
183Ling 2.6 Flash33.8%estimated ± 5.3 pp, low confidence
184Llama 3.3 70B33.8%estimated ± 5.3 pp, low confidence
185Llama 4 Maverick33.8%estimated ± 5.3 pp, low confidence
186Llama 4 Scout33.8%estimated ± 5.3 pp, low confidence
187Mercury 2.533.8%estimated ± 5.3 pp, low confidence
188Mistral Large 333.8%estimated ± 5.3 pp, low confidence
189Mistral Small 3.133.8%estimated ± 5.3 pp, low confidence
190Mistral Small 433.8%estimated ± 5.3 pp, low confidence
191Nemotron 3 Nano 30B33.8%estimated ± 5.3 pp, low confidence
192Nemotron 3 Nano Omni 30B A3B33.8%estimated ± 5.3 pp, low confidence
193North Mini Code33.8%estimated ± 5.3 pp, low confidence
194Solar Pro 333.8%estimated ± 5.3 pp, low confidence
195Trinity-Large-Preview33.8%estimated ± 5.3 pp, low confidence
196Trinity-Large-Thinking33.8%estimated ± 5.3 pp, low confidence
197Ultravox v0.6 Llama 3.3 70B33.8%estimated ± 5.3 pp, low confidence
198Ornith-1.5-35B-A3B33.4%measured
199o332.9%estimated ± 5.5 pp, low confidence
200LLaDA2.2-flash32.4%estimated ± 5.5 pp, low confidence
201Ternary Bonsai 2 27B32.3%estimated ± 5.5 pp, low confidence
202Ornith-1.5-9B30.5%measured
203DeepSeek V3.230.5%estimated ± 5.5 pp, low confidence
204Agents-A1-4B29.5%estimated ± 5.5 pp, low confidence
205GLM-4.627.8%estimated ± 5.5 pp, low confidence
206GPT-OSS 120B26.8%estimated ± 4.1 pp, low confidence
207MiMo-V2.5-Pro26.8%estimated ± 4.1 pp, low confidence
208GPT-OSS 20B26.8%estimated ± 4.1 pp, low confidence
209Grok Code Fast 126.2%estimated ± 5.5 pp, low confidence
210Grok 425.1%estimated ± 5.5 pp, low confidence
211Qwen3 Max24.3%estimated ± 5.5 pp, low confidence
212Claude 4 Opus Thinking23.1%estimated ± 5.5 pp, low confidence
213LFM3-1.7B13.9%measured

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General