benchgap
Agentic · tools

WideResearch leaderboard

As of 2026-11-21, the highest measured score on WideResearch is 83.9% by Hy4 preview. 187 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Muse Wonder Full86.8%estimated ± 1.2 pp, low confidence
2GLM-5.3-Flash85.9%estimated ± 3.9 pp, low confidence
3Inkling 2 Small85.7%estimated ± 1.2 pp, low confidence
4Muse Wonder85.7%estimated ± 1.2 pp, low confidence
5Claude Fable 5.185.6%estimated ± 3.9 pp, low confidence
6Claude Opus 5.585.6%estimated ± 3.9 pp, low confidence
7Claude Sonnet 5.585.6%estimated ± 3.9 pp, low confidence
8Grok 4.785.2%estimated ± 7.8 pp, low confidence
9Qwen3.8 Max Preview84.8%estimated ± 7.8 pp, low confidence
10LongCat-2.584.6%estimated ± 1.2 pp, low confidence
11Gemini 4 Argon84.5%estimated ± 7.8 pp, low confidence
12Muse Spark 1.384.5%estimated ± 1.2 pp, low confidence
13Grok 4.684.5%estimated ± 7.8 pp, low confidence
14GPT-6.1 Sol84.3%estimated ± 7.8 pp, low confidence
15Ultramarine 1 Flash84.1%estimated ± 1.2 pp, low confidence
16Hy4 preview83.9%measured
17MiMo-V2.6-Pro83.9%estimated ± 1.2 pp, low confidence
18Muse Spark Open83.8%estimated ± 1.2 pp, low confidence
19MiMo-V2.6-Flash83.7%estimated ± 1.2 pp, medium confidence
20GPT-6 Sol83.5%estimated ± 7.8 pp, low confidence
21Ling 3.1 Flash83.3%measured
22Step 5 Preview83.2%estimated ± 1.2 pp, medium confidence
23Kimi K3.1 Pro83.1%estimated ± 3.5 pp, high confidence
24Hy483.1%estimated ± 4.8 pp, low confidence
25DeepSeek V4.1 Flash 101783.1%estimated ± 1.6 pp, low confidence
26Muse Spark 1.283.1%estimated ± 7.8 pp, low confidence
27Fugu Cyber83.1%estimated ± 1.6 pp, medium confidence
28Gemini 3.8 Flash Cyber83.1%estimated ± 1.6 pp, medium confidence
29GLM-5.383.0%estimated ± 1.6 pp, medium confidence
30GPT-5.6 Sol83.0%estimated ± 1.6 pp, medium confidence
31Claude Mythos 583.0%estimated ± 1.6 pp, medium confidence
32DeepSeek V4 Pro83.0%estimated ± 1.6 pp, medium confidence
33Gemini 3.5 Flash Cyber83.0%estimated ± 1.6 pp, medium confidence
34Claude Mythos Preview83.0%estimated ± 1.6 pp, medium confidence
35MiniMax M3.1 Flash83.0%estimated ± 1.6 pp, medium confidence
36GPT-5.6 Terra83.0%estimated ± 1.6 pp, medium confidence
37GPT-6 Astra82.9%estimated ± 3.5 pp, high confidence
38GPT-5.6 Luna82.8%estimated ± 1.6 pp, medium confidence
39DeepSeek V4 Flash 073182.7%estimated ± 1.6 pp, medium confidence
40Gemini 3.8 Flash82.7%estimated ± 7.8 pp, low confidence
41GPT-6 Luna82.7%estimated ± 7.8 pp, low confidence
42Claude Opus 582.6%estimated ± 3.5 pp, high confidence
43Qwen3.8-Flash-Next82.5%estimated ± 1.2 pp, medium confidence
44GPT-5.5 Pro82.3%estimated ± 3.5 pp, high confidence
45Muse Spark 1.182.3%estimated ± 1.2 pp, medium confidence
46Gemini 3.7 Flash82.2%estimated ± 7.8 pp, low confidence
47GLM-5.182.1%estimated ± 1.6 pp, medium confidence
48Grok 4.582.1%estimated ± 7.8 pp, low confidence
49GPT-5.4 Pro82.0%estimated ± 3.5 pp, high confidence
50Atria Dawn Preview81.9%measured
51Qwen3.8 Max81.9%measured
52Kimi K381.9%estimated ± 1.2 pp, medium confidence
53Claude Haiku 5.581.9%estimated ± 4.8 pp, low confidence
54Claude Fable 581.4%estimated ± 2.1 pp, medium confidence
55Claude Opus 4.881.4%estimated ± 2.1 pp, medium confidence
56Gemini 3.6 Flash81.4%estimated ± 2.1 pp, medium confidence
57Holo3-35B-A3B81.4%estimated ± 2.1 pp, medium confidence
58Claude Sonnet 581.4%estimated ± 2.1 pp, medium confidence
59Holo3-122B-A10B81.4%estimated ± 2.1 pp, medium confidence
60Gemini 3.5 Flash81.4%estimated ± 2.1 pp, medium confidence
61UI-Mate-27B81.4%estimated ± 2.1 pp, medium confidence
62Gemini 3.5 Flash-Lite81.4%estimated ± 2.1 pp, medium confidence
63Qwen3.7 Plus81.4%estimated ± 2.1 pp, medium confidence
64GPT-5.4 mini81.4%estimated ± 2.1 pp, medium confidence
65MiniMax M381.3%estimated ± 2.1 pp, medium confidence
66Inkling 2 Mini81.2%estimated ± 1.2 pp, medium confidence
67ERNIE 5.5 Flash Preview81.2%estimated ± 3.5 pp, high confidence
68Apodex 1.180.9%estimated ± 4.8 pp, medium confidence
69Kimi K2.680.8%measured
70Ornith-1.5-397B80.8%measured
71Claude Opus 4.780.4%estimated ± 1.2 pp, medium confidence
72Hy4 flash80.2%estimated ± 4.8 pp, medium confidence
73Mistral Large 479.9%estimated ± 3.5 pp, high confidence
74GLM-5.279.8%estimated ± 4.8 pp, medium confidence
75GPT-5.579.7%estimated ± 1.2 pp, medium confidence
76Apodex 1.1 Mini79.1%estimated ± 7.8 pp, low confidence
77Quasar 438B79.0%estimated ± 7.8 pp, low confidence
78Qwen3.7 Max78.9%estimated ± 4.8 pp, medium confidence
79MiMo-V2.5-Pro78.9%estimated ± 5.9 pp, low confidence
80dots3-note Preview78.9%measured
81GPT-5.478.8%estimated ± 1.2 pp, medium confidence
82Sage-A1.578.6%estimated ± 3.5 pp, high confidence
83Ling 3.0 Flash VL78.5%estimated ± 7.8 pp, low confidence
84Claude Sonnet 4.678.4%estimated ± 1.2 pp, medium confidence
85Claude Opus 4.678.4%estimated ± 1.2 pp, medium confidence
86MiniMax M2.778.2%estimated ± 5.6 pp, low confidence
87Ling 3.0 Flash Sante77.9%estimated ± 3.5 pp, high confidence
88GPT-5.277.8%estimated ± 1.2 pp, medium confidence
89GPT-5.3 Codex77.7%estimated ± 1.2 pp, medium confidence
90Qwen3.8-27B77.6%estimated ± 1.2 pp, medium confidence
91Gemini 3.1 Pro77.3%estimated ± 5.9 pp, low confidence
92Nemotron 3.5 Super VL 120B A12B76.8%estimated ± 3.5 pp, high confidence
93Nemotron 3.5 Super VL 120B A12B NVFP476.6%estimated ± 3.5 pp, high confidence
94Inkling-Small76.5%estimated ± 3.5 pp, high confidence
95Qwen3.6-27B76.5%estimated ± 5.9 pp, low confidence
96Claude Opus 4.576.4%measured
97Inkling76.4%estimated ± 3.5 pp, high confidence
98Claude Sonnet 4.576.4%estimated ± 1.2 pp, medium confidence
99Hy376.3%estimated ± 7.8 pp, low confidence
100Beam 1.0 501B-A23B76.2%estimated ± 3.5 pp, high confidence
101Beam 1.0 501B-A23B FP876.2%estimated ± 3.5 pp, high confidence
102GPT-5.1-Codex76.1%estimated ± 1.2 pp, medium confidence
103UI-Mate-9B76.0%estimated ± 2.1 pp, medium confidence
104GPT-5.2-Codex76.0%estimated ± 1.2 pp, medium confidence
105Step 3.7 Flash75.8%estimated ± 3.5 pp, high confidence
106Beam 1.0 501B-A23B NVFP475.6%estimated ± 3.5 pp, high confidence
107Agents-A175.6%estimated ± 3.5 pp, high confidence
108Claude 4.1 Opus75.1%estimated ± 1.2 pp, medium confidence
109Qwen3.5 Plus74.4%estimated ± 1.2 pp, medium confidence
110Claude 4 Sonnet74.4%estimated ± 1.2 pp, medium confidence
111Qwen3.6 Plus74.3%measured
112Qwen3.5 397B74.0%measured
113Claude Haiku 4.573.8%estimated ± 1.2 pp, medium confidence
114Ling 3.0 Flash73.6%measured
115Qwen3.5-122B-A10B73.3%estimated ± 2.1 pp, low confidence
116Qwen3.5-27B73.3%estimated ± 2.1 pp, low confidence
117Qwen3.5-35B-A3B73.3%estimated ± 2.1 pp, low confidence
118GPT-5.4 nano73.3%estimated ± 2.1 pp, low confidence
119A.X K273.1%estimated ± 7.8 pp, low confidence
120Muse Glimmer 30B73.0%estimated ± 1.3 pp, low confidence
121Gemini 3 Flash72.8%estimated ± 1.2 pp, medium confidence
122Gemini 3 Pro72.8%estimated ± 1.2 pp, medium confidence
123Grok Build 0.172.8%estimated ± 5.9 pp, low confidence
124Nemotron 3.5 Bolt 74B-A9B BF1672.7%estimated ± 3.5 pp, high confidence
125Kimi K2.572.7%measured
126GPT-572.2%estimated ± 1.2 pp, low confidence
127Kimi K2.7 Code71.6%estimated ± 6.7 pp, low confidence
128Agents-A1-4B70.9%estimated ± 3.5 pp, high confidence
129Nemotron 3.5 Bolt 74B-A9B70.9%estimated ± 3.5 pp, high confidence
130Grok 4.1 Fast70.9%estimated ± 5.9 pp, low confidence
131Ripple-65B Preview70.9%estimated ± 6.7 pp, low confidence
132Solar Open 270.9%estimated ± 6.7 pp, low confidence
133LLaDA2.2-flash70.9%estimated ± 6.7 pp, low confidence
134MiMo-V2.570.4%estimated ± 5.9 pp, low confidence
135Muse Spark70.1%estimated ± 1.6 pp, medium confidence
136GLM-569.8%measured
137Laguna S 2.169.1%estimated ± 3.9 pp, medium confidence
138Ornith-1.5-35B-A3B67.8%measured
139Grok 4.366.2%estimated ± 5.9 pp, low confidence
140Qwen3 Max66.0%estimated ± 5.9 pp, low confidence
141K-EXAONE 2.065.7%estimated ± 7.8 pp, low confidence
142Grok 463.4%estimated ± 5.9 pp, low confidence
143Gemini 2.5 Pro62.8%estimated ± 5.9 pp, low confidence
144Spark-X2.6-4B62.4%estimated ± 3.5 pp, medium confidence
145GLM-4.761.6%estimated ± 3.5 pp, medium confidence
146GPT-5.161.2%estimated ± 5.9 pp, low confidence
147Qwen3.6-35B-A3B60.1%measured
148Sarvam 105B59.8%estimated ± 3.5 pp, medium confidence
149Solar Pro 459.6%estimated ± 3.5 pp, medium confidence
150Ornith-1.5-9B59.5%measured
151LongCat-Flash-Lite-Sparse59.1%estimated ± 3.5 pp, medium confidence
152MiniCPM5-2B57.2%estimated ± 7.8 pp, low confidence
153Mistral Medium 3.5 128B56.0%estimated ± 5.9 pp, low confidence
154Nemotron 3 Ultra55.9%estimated ± 3.5 pp, medium confidence
155Gemini 3.1 Flash-Lite54.2%estimated ± 5.9 pp, low confidence
156Grok 4.2054.0%estimated ± 5.9 pp, low confidence
157Hy3 Preview49.7%estimated ± 5.9 pp, low confidence
158Nemotron 3.5 Lightning 30B A3B NVFP449.4%estimated ± 3.5 pp, medium confidence
159LFM3-1.7B49.2%estimated ± 4.8 pp, low confidence
160MiMo-V2-Pro48.9%estimated ± 5.9 pp, low confidence
161Sarvam 30B48.2%estimated ± 3.5 pp, medium confidence
162MiMo-V2-Flash45.9%estimated ± 7.8 pp, low confidence
163Gemma 4 31B44.2%estimated ± 5.9 pp, low confidence
164Trinity-Large-Thinking34.6%estimated ± 5.9 pp, low confidence
165Granite 4.2 30B33.3%estimated ± 7.8 pp, low confidence
166Gemma 4 26B A4B30.3%estimated ± 7.8 pp, low confidence
167Ling 3.0 Tiny29.0%estimated ± 7.8 pp, low confidence
168GLM-5V-Turbo28.1%estimated ± 5.9 pp, low confidence
169GPT-OSS 120B24.0%estimated ± 5.9 pp, low confidence
170DeepSeek V3.223.9%estimated ± 5.9 pp, low confidence
171GPT-4.112.3%estimated ± 5.9 pp, low confidence
172Command A+8.8%estimated ± 7.8 pp, low confidence
173Celeris-10.0%estimated ± 7.8 pp, low confidence
174Claude 3.5 Haiku0.0%estimated ± 7.8 pp, low confidence
175DeepSeek V30.0%estimated ± 7.8 pp, low confidence
176DeepSeek V3 03240.0%estimated ± 7.8 pp, low confidence
177Gemma 3 27B0.0%estimated ± 7.8 pp, low confidence
178Gemma 4 12B0.0%estimated ± 7.8 pp, low confidence
179Gemma 4 E2B0.0%estimated ± 7.8 pp, low confidence
180Gemma 4 E4B0.0%estimated ± 7.8 pp, low confidence
181GPT-4.1 mini0.0%estimated ± 7.8 pp, low confidence
182GPT-4.1 nano0.0%estimated ± 7.8 pp, low confidence
183GPT-4o0.0%estimated ± 7.8 pp, low confidence
184GPT-4o mini0.0%estimated ± 7.8 pp, low confidence
185GPT-OSS 20B0.0%estimated ± 7.8 pp, low confidence
186Granite 4.2 3B0.0%estimated ± 7.8 pp, low confidence
187Granite 4.2 8B0.0%estimated ± 7.8 pp, low confidence
188K-Exaone0.0%estimated ± 7.8 pp, low confidence
189LFM2.5-2.6B0.0%estimated ± 7.8 pp, low confidence
190Ling 2.6 Flash0.0%estimated ± 7.8 pp, low confidence
191Llama 3.3 70B0.0%estimated ± 7.8 pp, low confidence
192Llama 4 Maverick0.0%estimated ± 7.8 pp, low confidence
193Llama 4 Scout0.0%estimated ± 7.8 pp, low confidence
194Mercury 2.50.0%estimated ± 7.8 pp, low confidence
195Mistral Large 30.0%estimated ± 7.8 pp, low confidence
196Mistral Small 3.10.0%estimated ± 7.8 pp, low confidence
197Mistral Small 40.0%estimated ± 7.8 pp, low confidence
198Nemotron 3 Nano 30B0.0%estimated ± 7.8 pp, low confidence
199Nemotron 3 Nano Omni 30B A3B0.0%estimated ± 7.8 pp, low confidence
200North Mini Code0.0%estimated ± 7.8 pp, low confidence
201Solar Pro 30.0%estimated ± 7.8 pp, low confidence
202Trinity-Large-Preview0.0%estimated ± 7.8 pp, low confidence
203Ultravox v0.6 Llama 3.3 70B0.0%estimated ± 7.8 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General