Agentic · tools

BrowseComp leaderboard

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

As of 2026-10-07, the highest measured score on BrowseComp is 92.5% by Atria Dawn Preview. 143 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Atria Dawn Preview	92.5%	measured
2	GPT-5.6 Sol	92.2%	measured
3	GPT-6 Astra	91.5%	measured
4	Gemini 4 Argon	91.5%	estimated ± 2.2 pp, high confidence
5	Claude Fable 5	91.2%	estimated ± 2.8 pp, high confidence
6	Kimi K3	91.2%	measured
7	Muse Spark 1.3	91.1%	estimated ± 2.2 pp, high confidence
8	Claude Opus 5	90.8%	measured
9	GPT-6.1 Sol	90.7%	estimated ± 1.8 pp, medium confidence
10	GPT-5.5 Pro	90.1%	measured
11	Gemini 3.8 Flash	89.4%	estimated ± 2.2 pp, high confidence
12	GPT-5.4 Pro	89.3%	measured
13	GLM-5.3-Flash	89.0%	estimated ± 4.7 pp, low confidence
14	Ling 3.1 Flash	88.9%	estimated ± 5.2 pp, low confidence
15	Step 5 Preview	88.7%	measured
16	Fugu Cyber	88.3%	estimated ± 5.2 pp, low confidence
17	Claude Mythos 5	88.0%	measured
18	Gemini 3.8 Flash Cyber	87.9%	estimated ± 5.2 pp, low confidence
19	GPT-5.6 Terra	87.5%	measured
20	GPT-6 Sol	87.0%	estimated ± 1.8 pp, medium confidence
21	Claude Sonnet 5.5	87.0%	estimated ± 6.4 pp, medium confidence
22	Ornith-1.5-397B	86.6%	measured
23	Claude Opus 5.5	86.3%	estimated ± 2.2 pp, high confidence
24	Gemini 3.5 Flash Cyber	86.2%	estimated ± 5.2 pp, low confidence
25	Qwen3.8 Max Preview	86.2%	estimated ± 3.4 pp, medium confidence
26	Gemini 3.7 Flash	86.1%	estimated ± 2.2 pp, high confidence
27	GLM-5.2	85.8%	estimated ± 3.4 pp, high confidence
28	MiMo-V2.6-Pro	85.8%	estimated ± 1.8 pp, medium confidence
29	Claude Opus 4.6 (Adaptive)	85.7%	estimated ± 3.4 pp, high confidence
30	Claude Mythos Preview	85.7%	estimated ± 1.8 pp, medium confidence
31	Gemini 3.1 Pro	85.6%	estimated ± 3.4 pp, high confidence
32	Apodex 1.1	85.6%	estimated ± 3.4 pp, high confidence
33	Apodex 1.1 Mini	85.6%	estimated ± 3.4 pp, high confidence
34	DeepSeek V4.1 Flash	85.1%	estimated ± 1.8 pp, medium confidence
35	GLM-5.3	85.0%	estimated ± 1.8 pp, medium confidence
36	Muse Spark 1.2	84.9%	estimated ± 9.0 pp, medium confidence
37	Claude Sonnet 5	84.7%	measured
38	Claude Haiku 5.5	84.6%	estimated ± 7.6 pp, medium confidence
39	Claude Fable 5.1	84.5%	estimated ± 2.2 pp, high confidence
40	Grok 4.6	84.5%	estimated ± 2.8 pp, high confidence
41	GPT-5.5	84.4%	measured
42	Claude Opus 4.8	84.3%	measured
43	Grok 4.5	84.2%	estimated ± 9.0 pp, medium confidence
44	GPT-6 Luna	84.0%	estimated ± 1.8 pp, medium confidence
45	Gemini 3.6 Flash	83.8%	estimated ± 3.1 pp, high confidence
46	Holo3-35B-A3B	83.8%	estimated ± 3.1 pp, high confidence
47	Qwen3.8-27B	83.8%	estimated ± 3.1 pp, high confidence
48	Holo3-122B-A10B	83.8%	estimated ± 3.1 pp, high confidence
49	Gemini 3.5 Flash	83.8%	estimated ± 3.1 pp, high confidence
50	UI-Mate-27B	83.8%	estimated ± 3.1 pp, high confidence
51	Gemini 3.5 Flash-Lite	83.8%	estimated ± 3.1 pp, high confidence
52	GPT-5.4 mini	83.7%	estimated ± 3.1 pp, high confidence
53	Claude Opus 4.6	83.7%	measured
54	MiniMax M3	83.5%	measured
55	DeepSeek V4 Pro 0813	83.4%	measured
56	Hy4 preview	83.3%	estimated ± 5.2 pp, low confidence
57	dots3-note Preview	83.3%	measured
58	GPT-5.6 Luna	83.3%	measured
59	Kimi K2.6	83.2%	measured
60	Qwen3.8-Flash-Next	82.8%	estimated ± 2.2 pp, high confidence
61	Qwen3.8 Max	82.8%	estimated ± 2.2 pp, high confidence
62	Claude Opus 4.7	82.8%	estimated ± 2.2 pp, high confidence
63	Claude Sonnet 4.6	82.8%	estimated ± 2.2 pp, high confidence
64	Qwen3.7 Plus	82.8%	estimated ± 2.2 pp, medium confidence
65	Mistral Large 4	82.7%	estimated ± 10.1 pp, low confidence
66	GPT-5.4	82.7%	measured
67	MiMo-V2.6-Flash	82.4%	estimated ± 1.8 pp, medium confidence
68	K-EXAONE 2.0	82.0%	estimated ± 9.4 pp, low confidence
69	Ornith-1.0-397B	81.6%	estimated ± 9.4 pp, low confidence
70	Qwen3.7 Max	81.1%	estimated ± 5.7 pp, medium confidence
71	Gemini 3 Pro	81.1%	estimated ± 5.7 pp, medium confidence
72	Muse Spark 1.1	80.9%	estimated ± 1.8 pp, low confidence
73	Gemini 3 Flash	80.7%	estimated ± 5.7 pp, medium confidence
74	Qwen3.6-27B	80.0%	estimated ± 5.7 pp, medium confidence
75	Quasar 438B	79.8%	estimated ± 9.0 pp, medium confidence
76	Kimi K2.7 Code	79.8%	estimated ± 8.4 pp, medium confidence
77	Claude Opus 4.5	79.4%	estimated ± 3.1 pp, high confidence
78	Claude Opus 4.7 (Adaptive)	79.3%	measured
79	Grok 4.7	79.1%	estimated ± 4.7 pp, low confidence
80	UI-Mate-9B	79.1%	estimated ± 3.1 pp, high confidence
81	Grok 4.3	78.7%	estimated ± 3.4 pp, high confidence
82	Muse Glimmer 30B	78.3%	estimated ± 3.1 pp, high confidence
83	Inkling-Small	77.4%	measured
84	Beam	77.4%	measured
85	Inkling	77.1%	measured
86	Ornith-1.0-35B	76.9%	estimated ± 9.4 pp, low confidence
87	GPT-5.2-Codex	76.5%	estimated ± 5.7 pp, medium confidence
88	Mercury 2.5	76.2%	estimated ± 8.2 pp, low confidence
89	Step 3.7 Flash	75.8%	measured
90	Agents-A1	75.5%	measured
91	Ling 3.0 Flash VL	75.3%	estimated ± 10.1 pp, low confidence
92	Hy3	75.1%	estimated ± 9.0 pp, medium confidence
93	GPT-5.3 Codex	74.3%	estimated ± 3.1 pp, high confidence
94	GLM-5	73.4%	estimated ± 3.4 pp, high confidence
95	DeepSeek V4 Flash 0731	73.2%	measured
96	Ling 3.0 Flash	72.2%	measured
97	Ornith-1.0-9B	72.2%	estimated ± 9.4 pp, low confidence
98	LFM2.5-2.6B	72.0%	estimated ± 9.4 pp, low confidence
99	Laguna S 2.1	71.7%	estimated ± 6.4 pp, medium confidence
100	GPT-5.1-Codex	71.2%	estimated ± 5.7 pp, medium confidence
101	Ling 3.0 Flash FP8	70.7%	estimated ± 9.0 pp, medium confidence
102	Grok Build 0.1	69.6%	estimated ± 5.7 pp, medium confidence
103	Claude 4.1 Opus	68.4%	estimated ± 6.2 pp, medium confidence
104	GLM-5.1	68.0%	measured
105	LLaDA2.2-mini	67.7%	estimated ± 9.4 pp, low confidence
106	Ornith-1.5-35B-A3B	67.6%	measured
107	Qwen3.6 Plus	66.9%	estimated ± 4.4 pp, high confidence
108	Agents-A1-4B	66.8%	measured
109	GLM-5-Turbo	66.6%	estimated ± 9.4 pp, low confidence
110	A.X K2	66.2%	estimated ± 10.1 pp, low confidence
111	Qwen3.5 Plus	66.2%	estimated ± 6.2 pp, medium confidence
112	GPT-5.2	65.8%	measured
113	Claude Sonnet 4.5	65.1%	estimated ± 3.1 pp, high confidence
114	Grok 4.1 Fast	64.6%	estimated ± 5.7 pp, medium confidence
115	Claude Haiku 4.5	64.5%	estimated ± 6.2 pp, medium confidence
116	Solar Open 2	63.9%	estimated ± 8.4 pp, medium confidence
117	Qwen3.5-122B-A10B	63.8%	measured
118	MiMo-V2.5	63.6%	estimated ± 5.7 pp, medium confidence
119	GPT-5.4 nano	62.8%	estimated ± 3.1 pp, medium confidence
120	Qwen3.5 397B	62.0%	measured
121	Qwen3.5-27B	61.0%	measured
122	Qwen3.5-35B-A3B	61.0%	measured
123	Kimi K2.5	60.6%	measured
124	Kimi K2.5 (Reasoning)	60.6%	measured
125	Qwen3 Max	59.8%	estimated ± 5.7 pp, medium confidence
126	GPT-5 (high)	59.6%	estimated ± 6.2 pp, low confidence
127	Grok 4	59.2%	estimated ± 5.7 pp, medium confidence
128	Gemini 2.5 Pro	59.1%	estimated ± 5.7 pp, medium confidence
129	LLaDA2.2-flash	59.0%	estimated ± 8.4 pp, medium confidence
130	GPT-5.1	58.9%	estimated ± 5.7 pp, medium confidence
131	Qwen3.6-35B-A3B	58.9%	estimated ± 4.4 pp, medium confidence
132	Claude 4 Sonnet	58.8%	estimated ± 5.7 pp, medium confidence
133	Mistral Medium 3.5 128B	58.7%	estimated ± 5.7 pp, medium confidence
134	Gemini 3.1 Flash-Lite	58.7%	estimated ± 5.7 pp, medium confidence
135	Grok 4.20	58.7%	estimated ± 5.7 pp, medium confidence
136	Hy3 Preview	58.7%	estimated ± 5.7 pp, medium confidence
137	MiMo-V2-Pro	58.7%	estimated ± 5.7 pp, medium confidence
138	Gemma 4 31B	58.6%	estimated ± 5.7 pp, medium confidence
139	Trinity-Large-Thinking	58.6%	estimated ± 5.7 pp, medium confidence
140	GLM-5V-Turbo	58.6%	estimated ± 5.7 pp, medium confidence
141	DeepSeek V3.2	58.6%	estimated ± 5.7 pp, medium confidence
142	GPT-4.1	58.6%	estimated ± 5.7 pp, low confidence
143	MiMo-V2-Omni	57.6%	estimated ± 9.4 pp, low confidence
144	Muse Spark	56.7%	estimated ± 5.2 pp, low confidence
145	Ornith-1.5-9B	56.4%	measured
146	MiniMax M2.7	55.2%	estimated ± 3.4 pp, medium confidence
147	GLM-4.7	52.0%	measured
148	Solar Pro 4	49.2%	measured
149	LongCat-Flash-Lite-Sparse	48.6%	measured
150	Grok 4.1	46.0%	estimated ± 7.1 pp, low confidence
151	MiniCPM5-2B	44.6%	estimated ± 10.1 pp, low confidence
152	Nemotron 3 Ultra	44.4%	measured
153	Nemotron 3.5 Lightning 30B A3B NVFP4	36.8%	measured
154	MiMo-V2-Flash	32.9%	estimated ± 10.1 pp, low confidence
155	Granite 4.2 8B	28.5%	estimated ± 9.0 pp, low confidence
156	Command A+	28.0%	estimated ± 9.0 pp, low confidence
157	Granite 4.2 30B	22.2%	estimated ± 10.1 pp, low confidence
158	Mistral Large 3	20.7%	estimated ± 9.0 pp, low confidence
159	Gemma 4 26B A4B	19.8%	estimated ± 10.1 pp, low confidence
160	Ling 3.0 Tiny	18.9%	estimated ± 10.1 pp, low confidence
161	Mistral Small 4	13.3%	estimated ± 9.0 pp, low confidence
162	Mistral Small 4 (Reasoning)	13.3%	estimated ± 9.0 pp, low confidence
163	Trinity-Large-Preview	11.0%	estimated ± 9.0 pp, low confidence
164	Nemotron 3 Nano 30B	9.6%	estimated ± 9.0 pp, low confidence
165	DeepSeek V3	7.7%	estimated ± 9.0 pp, low confidence
166	Celeris-1	6.4%	estimated ± 9.0 pp, low confidence
167	Llama 4 Maverick	6.2%	estimated ± 9.0 pp, low confidence
168	Llama 4 Scout	5.5%	estimated ± 9.0 pp, low confidence
169	GPT-OSS 120B	1.6%	estimated ± 3.4 pp, medium confidence
170	Gemma 3 27B	1.5%	estimated ± 9.0 pp, low confidence
171	MiMo-V2.5-Pro	0.6%	estimated ± 3.4 pp, medium confidence
172	Nemotron 3 Super 100B	0.2%	estimated ± 3.4 pp, medium confidence
173	GPT-OSS 20B	0.0%	estimated ± 3.4 pp, medium confidence
174	DeepSeek V3 0324	0.0%	estimated ± 10.1 pp, low confidence
175	Gemma 4 12B	0.0%	estimated ± 10.1 pp, low confidence
176	Gemma 4 E2B	0.0%	estimated ± 10.1 pp, low confidence
177	Gemma 4 E4B	0.0%	estimated ± 10.1 pp, low confidence
178	GPT-4.1 mini	0.0%	estimated ± 10.1 pp, low confidence
179	GPT-4.1 nano	0.0%	estimated ± 10.1 pp, low confidence
180	GPT-4o	0.0%	estimated ± 10.1 pp, low confidence
181	GPT-4o mini	0.0%	estimated ± 10.1 pp, low confidence
182	Granite 4.2 3B	0.0%	estimated ± 10.1 pp, low confidence
183	K-Exaone	0.0%	estimated ± 10.1 pp, low confidence
184	Ling 2.6 Flash	0.0%	estimated ± 10.1 pp, low confidence
185	Nemotron 3 Nano Omni 30B A3B	0.0%	estimated ± 10.1 pp, low confidence
186	North Mini Code	0.0%	estimated ± 10.1 pp, low confidence
187	Solar Pro 3	0.0%	estimated ± 10.1 pp, low confidence
188	Ultravox v0.6 Llama 3.3 70B	0.0%	estimated ± 10.1 pp, low confidence

BrowseComp leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General