Knowledge & reasoning

LABBench2 leaderboard

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

As of 2026-10-07, the highest measured score on LABBench2 is 88.8% by Gemini 4 Argon. 205 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Gemini 4 Argon	88.8%	measured
2	Gemini 3.8 Flash	86.2%	measured
3	Claude Sonnet 5.5	86.2%	estimated ± 2.4 pp, medium confidence
4	Claude Fable 5	85.6%	estimated ± 1.3 pp, medium confidence
5	GPT-6 Astra	85.6%	estimated ± 1.3 pp, medium confidence
6	Claude Opus 5	84.2%	measured
7	Claude Fable 5.1	84.1%	estimated ± 1.3 pp, medium confidence
8	Claude Opus 5.5	84.1%	estimated ± 1.3 pp, medium confidence
9	MiMo-V2.6-Pro	83.8%	estimated ± 2.4 pp, medium confidence
10	Gemini 3.1 Pro	83.6%	estimated ± 1.8 pp, low confidence
11	Claude Opus 4.7	83.0%	estimated ± 1.8 pp, low confidence
12	Qwen3.7 Max	82.7%	estimated ± 1.8 pp, low confidence
13	Muse Spark 1.3	82.7%	estimated ± 1.8 pp, low confidence
14	GPT-6.1 Sol	82.6%	estimated ± 1.3 pp, medium confidence
15	Step 5 Preview	82.6%	estimated ± 2.4 pp, medium confidence
16	Muse Spark 1.1	82.4%	estimated ± 1.8 pp, low confidence
17	Gemini 3 Flash	82.3%	estimated ± 1.8 pp, low confidence
18	Qwen3.8 Max	82.3%	estimated ± 1.8 pp, low confidence
19	Muse Spark 1.2	82.2%	estimated ± 1.8 pp, low confidence
20	Gemini 3.7 Flash	82.1%	measured
21	GPT-5.6 Sol	82.1%	measured
22	Qwen3.6 Plus	81.8%	estimated ± 1.8 pp, low confidence
23	Kimi K2.6	81.8%	estimated ± 1.8 pp, low confidence
24	Claude Haiku 5.5	81.7%	estimated ± 2.4 pp, medium confidence
25	Muse Spark	81.6%	estimated ± 1.8 pp, low confidence
26	Qwen3.8 Max Preview	81.5%	estimated ± 1.8 pp, low confidence
27	GLM-5.1	81.4%	estimated ± 1.8 pp, low confidence
28	GLM-5.3	81.3%	estimated ± 1.8 pp, low confidence
29	GLM-5.2	81.3%	estimated ± 1.8 pp, low confidence
30	GPT-5.6 Terra	81.2%	measured
31	Grok 4.7	81.1%	estimated ± 2.4 pp, medium confidence
32	Grok 4.20	81.1%	estimated ± 1.8 pp, low confidence
33	Inkling	81.1%	estimated ± 1.8 pp, low confidence
34	Gemini 3.1 Flash-Lite	81.0%	estimated ± 1.8 pp, low confidence
35	GPT-6 Sol	81.0%	estimated ± 1.3 pp, medium confidence
36	GLM-5.3-Flash	80.9%	estimated ± 1.8 pp, low confidence
37	Gemini 3.5 Flash-Lite	80.8%	estimated ± 1.8 pp, low confidence
38	Grok 4.3	80.8%	estimated ± 1.8 pp, low confidence
39	Nemotron 3 Ultra	80.8%	estimated ± 1.8 pp, low confidence
40	Qwen3.8-Flash-Next	80.7%	estimated ± 1.8 pp, low confidence
41	dots3-note Preview	80.7%	estimated ± 2.5 pp, low confidence
42	Claude Sonnet 4.5	80.7%	estimated ± 2.5 pp, low confidence
43	Gemini 3 Pro Deep Think	80.7%	estimated ± 2.5 pp, low confidence
44	GPT-5.5	80.2%	estimated ± 1.3 pp, low confidence
45	GPT-5.5 Pro	80.2%	estimated ± 1.3 pp, low confidence
46	MiMo-V2.5-Pro	80.1%	estimated ± 1.8 pp, low confidence
47	Claude Sonnet 5	80.1%	measured
48	Qwen3.8-27B	79.9%	estimated ± 1.8 pp, low confidence
49	MiniMax M3	79.9%	estimated ± 1.8 pp, low confidence
50	Qwen3.5 Flash	79.8%	estimated ± 1.8 pp, low confidence
51	Ling 3.1 Flash	79.5%	estimated ± 2.4 pp, low confidence
52	DeepSeek V4.1 Flash	79.5%	estimated ± 2.4 pp, low confidence
53	GPT-5.3 Codex	79.4%	estimated ± 1.8 pp, low confidence
54	GPT-5.4 Pro	79.3%	estimated ± 1.3 pp, low confidence
55	Kimi K3	79.3%	estimated ± 1.3 pp, low confidence
56	Claude Opus 4.7 (Adaptive)	79.3%	estimated ± 1.8 pp, low confidence
57	MiMo-V2.5	79.1%	estimated ± 1.8 pp, low confidence
58	GLM-4.7	79.0%	estimated ± 1.8 pp, low confidence
59	GLM-4.6	78.7%	estimated ± 1.8 pp, low confidence
60	Ling 3.0 Flash	78.6%	estimated ± 1.8 pp, low confidence
61	GLM-4.5	78.1%	estimated ± 1.8 pp, low confidence
62	GPT-5.4	78.0%	estimated ± 1.3 pp, low confidence
63	MiMo-V2.6-Flash	77.7%	estimated ± 2.4 pp, low confidence
64	MiniMax M2.7	77.7%	estimated ± 1.8 pp, low confidence
65	Mistral Large 4	77.7%	estimated ± 2.4 pp, low confidence
66	Qwen3.7 Plus	77.1%	estimated ± 1.8 pp, low confidence
67	GPT-5.2-Codex	76.9%	estimated ± 1.8 pp, low confidence
68	Claude Opus 4.6 (Adaptive)	76.8%	estimated ± 1.3 pp, low confidence
69	Claude Haiku 4.5	76.7%	estimated ± 1.8 pp, low confidence
70	Hy3	76.6%	estimated ± 1.8 pp, low confidence
71	Hy3 Preview	76.6%	estimated ± 1.8 pp, low confidence
72	Kimi K2.7 Code	76.4%	estimated ± 1.8 pp, low confidence
73	Claude Opus 4.8	75.9%	estimated ± 1.3 pp, low confidence
74	Gemini 3.5 Flash	75.9%	estimated ± 1.3 pp, low confidence
75	Solar Pro 4	75.7%	estimated ± 1.8 pp, low confidence
76	Qwen 3.6 Max (preview)	75.2%	estimated ± 1.8 pp, low confidence
77	Mistral Medium 3.5 128B	74.6%	estimated ± 1.8 pp, low confidence
78	Kimi K2.5	73.8%	estimated ± 1.8 pp, low confidence
79	Kimi K2.5 (Reasoning)	73.8%	estimated ± 1.8 pp, low confidence
80	Gemini 3.6 Flash	73.6%	estimated ± 1.3 pp, low confidence
81	Grok 4	73.5%	estimated ± 1.8 pp, low confidence
82	MiMo-V2-Pro	72.5%	estimated ± 1.8 pp, low confidence
83	Apodex 1.1	71.6%	estimated ± 1.8 pp, low confidence
84	Apodex 1.1 Mini	71.6%	estimated ± 1.8 pp, low confidence
85	DeepSeek V4 Pro 0813	71.4%	estimated ± 1.3 pp, low confidence
86	Ling 3.0 Flash VL	71.3%	estimated ± 1.8 pp, low confidence
87	Qwen3.5 397B	71.2%	estimated ± 1.8 pp, low confidence
88	Qwen3.5 397B (Reasoning)	71.2%	estimated ± 1.8 pp, low confidence
89	GPT-5.1-Codex	71.0%	estimated ± 1.8 pp, low confidence
90	GPT-5.1-Codex-Max	71.0%	estimated ± 1.8 pp, low confidence
91	Qwen3.5-27B	70.8%	estimated ± 1.8 pp, low confidence
92	A.X K2	70.6%	estimated ± 1.8 pp, low confidence
93	Gemma 4 31B	70.6%	estimated ± 1.8 pp, low confidence
94	Qwen3.5-122B-A10B	70.6%	estimated ± 1.8 pp, low confidence
95	Laguna XS.2	70.6%	estimated ± 1.8 pp, low confidence
96	Laguna M.1	70.4%	estimated ± 1.8 pp, low confidence
97	Ling 3.0 Flash FP8	70.3%	estimated ± 1.8 pp, low confidence
98	GPT-5 (high)	70.2%	estimated ± 1.8 pp, low confidence
99	Grok 4.1 Fast (Reasoning)	70.0%	estimated ± 1.8 pp, low confidence
100	DeepSeek V4 Flash 0731	69.6%	estimated ± 1.3 pp, low confidence
101	GLM-5-Turbo	69.2%	estimated ± 1.8 pp, low confidence
102	Grok 4 Fast (Reasoning)	69.2%	estimated ± 1.8 pp, low confidence
103	o3-pro	68.9%	estimated ± 1.8 pp, low confidence
104	Qwen3.5-35B-A3B	68.9%	estimated ± 1.8 pp, low confidence
105	Gemini 2.5 Pro	68.8%	estimated ± 1.8 pp, low confidence
106	GPT-5 (medium)	68.5%	estimated ± 1.8 pp, low confidence
107	Qwen3.6-27B	68.5%	estimated ± 1.8 pp, low confidence
108	Qwen3.6-35B-A3B	68.3%	estimated ± 1.8 pp, low confidence
109	Claude Opus 4.6	68.2%	estimated ± 1.8 pp, low confidence
110	Mercury 2.5	67.8%	estimated ± 2.4 pp, low confidence
111	GPT-5.6 Luna	67.6%	estimated ± 1.3 pp, low confidence
112	Muse Glimmer 30B	67.5%	estimated ± 1.8 pp, low confidence
113	K-EXAONE 2.0	66.7%	estimated ± 1.8 pp, low confidence
114	MiMo-V2-Omni	66.5%	estimated ± 1.8 pp, low confidence
115	o3	66.4%	estimated ± 1.8 pp, low confidence
116	Grok 4.6	65.7%	estimated ± 1.3 pp, low confidence
117	GLM-5	65.4%	estimated ± 1.8 pp, low confidence
118	GPT-6 Luna	65.1%	estimated ± 1.3 pp, low confidence
119	DeepSeek-R1	64.5%	estimated ± 1.8 pp, low confidence
120	Claude Opus 4.5	64.1%	estimated ± 1.8 pp, low confidence
121	GPT-4 Turbo	64.1%	estimated ± 2.4 pp, low confidence
122	GPT-5.2	64.1%	estimated ± 1.3 pp, low confidence
123	Claude 4.1 Opus Thinking	64.0%	estimated ± 1.8 pp, low confidence
124	GLM-5V-Turbo	64.0%	estimated ± 1.8 pp, low confidence
125	Step 3.7 Flash	64.0%	estimated ± 1.8 pp, low confidence
126	Claude Sonnet 4.6	63.7%	estimated ± 1.3 pp, low confidence
127	Grok 4.5	63.1%	estimated ± 1.3 pp, low confidence
128	Nemotron 3 Super 100B	62.8%	estimated ± 1.8 pp, low confidence
129	Gemma 4 26B A4B	61.7%	estimated ± 1.8 pp, low confidence
130	K-Exaone	60.6%	estimated ± 1.8 pp, low confidence
131	GPT-OSS 120B	60.5%	estimated ± 1.8 pp, low confidence
132	DeepSeek V3.1 (Reasoning)	60.1%	estimated ± 1.8 pp, low confidence
133	Inkling-Small	59.7%	estimated ± 1.3 pp, low confidence
134	Mistral Small 4	58.8%	estimated ± 1.8 pp, low confidence
135	Mistral Small 4 (Reasoning)	58.8%	estimated ± 1.8 pp, low confidence
136	Kimi K2	58.5%	estimated ± 1.8 pp, low confidence
137	o1-preview	58.3%	estimated ± 1.8 pp, low confidence
138	Qwen3 Max	58.2%	estimated ± 1.8 pp, low confidence
139	Command A+	57.8%	estimated ± 1.8 pp, low confidence
140	Nemotron 3 Nano 30B	57.4%	estimated ± 1.8 pp, low confidence
141	North Mini Code	57.4%	estimated ± 1.8 pp, low confidence
142	Gemma 4 12B	56.9%	estimated ± 1.8 pp, low confidence
143	Trinity-Large-Preview	56.7%	estimated ± 1.8 pp, low confidence
144	Trinity-Large-Thinking	56.7%	estimated ± 1.8 pp, low confidence
145	DeepSeek V3.2	56.6%	estimated ± 1.8 pp, low confidence
146	o3-mini	56.3%	estimated ± 1.8 pp, low confidence
147	o1	56.1%	estimated ± 1.8 pp, low confidence
148	Nemotron 3.5 Lightning 30B A3B NVFP4	55.7%	estimated ± 1.8 pp, low confidence
149	Sarvam 105B	55.1%	estimated ± 1.8 pp, low confidence
150	DeepSeek V3.1	54.7%	estimated ± 1.8 pp, low confidence
151	Ling 3.0 Tiny	54.6%	estimated ± 1.8 pp, low confidence
152	GLM-4.5-Air	54.5%	estimated ± 1.8 pp, low confidence
153	Quasar 438B	54.4%	estimated ± 1.8 pp, low confidence
154	Nemotron Ultra 253B	53.9%	estimated ± 1.8 pp, low confidence
155	Grok Code Fast 1	53.8%	estimated ± 1.8 pp, low confidence
156	Qwen3-Omni-30B-A3B-Thinking	53.7%	estimated ± 1.8 pp, low confidence
157	Solar Pro 3	53.4%	estimated ± 1.8 pp, low confidence
158	Claude Opus 4.5 Thinking	51.6%	estimated ± 1.3 pp, low confidence
159	MiniCPM5-2B	50.9%	estimated ± 1.8 pp, low confidence
160	GPT-OSS 20B	49.4%	estimated ± 1.8 pp, low confidence
161	Claude 4 Sonnet	48.8%	estimated ± 1.8 pp, low confidence
162	Gemini 2.5 Flash	48.8%	estimated ± 1.8 pp, low confidence
163	Mistral Large 3	48.5%	estimated ± 1.8 pp, low confidence
164	Llama 4 Maverick	47.5%	estimated ± 1.8 pp, low confidence
165	GPT-4.1	47.0%	estimated ± 1.8 pp, low confidence
166	GPT-4.1 mini	46.8%	estimated ± 1.8 pp, low confidence
167	MiMo-V2-Flash	46.0%	estimated ± 1.8 pp, low confidence
168	DeepSeek V3 0324	45.9%	estimated ± 1.8 pp, low confidence
169	Granite 4.2 30B	44.7%	estimated ± 1.8 pp, low confidence
170	Grok 4.1 Fast	44.0%	estimated ± 1.8 pp, low confidence
171	Sarvam 30B	43.6%	estimated ± 1.8 pp, low confidence
172	Celeris-1	43.4%	estimated ± 1.8 pp, low confidence
173	Granite 4.2 8B	43.4%	estimated ± 1.8 pp, low confidence
174	Exaone 4.0 32B	43.1%	estimated ± 1.8 pp, low confidence
175	Qwen3-Omni-30B-A3B-Instruct	42.3%	estimated ± 1.8 pp, low confidence
176	DeepSeek R1 Distill Qwen 32B	41.8%	estimated ± 1.8 pp, low confidence
177	Gemini 3 Pro	41.6%	estimated ± 1.3 pp, low confidence
178	Ling 2.6 Flash	39.7%	estimated ± 1.8 pp, low confidence
179	Gemini 1.5 Pro	39.3%	estimated ± 1.8 pp, low confidence
180	Llama 4 Scout	39.1%	estimated ± 1.8 pp, low confidence
181	Mistral Medium 3	38.3%	estimated ± 1.8 pp, low confidence
182	Gemma 4 E4B	38.1%	estimated ± 1.8 pp, low confidence
183	Phi-4	38.0%	estimated ± 1.8 pp, low confidence
184	GPT-5.1	37.4%	estimated ± 1.3 pp, low confidence
185	Solar Pro 2	36.7%	estimated ± 1.8 pp, low confidence
186	Granite 4.2 3B	36.5%	estimated ± 1.8 pp, low confidence
187	LFM2.5-2.6B	36.4%	estimated ± 1.8 pp, low confidence
188	DeepSeek V3	36.4%	estimated ± 1.8 pp, low confidence
189	GPT-4o	35.1%	estimated ± 1.8 pp, low confidence
190	Llama 3.1 405B	32.7%	estimated ± 1.8 pp, low confidence
191	LFM2.5-8B-A1B	32.5%	estimated ± 1.8 pp, low confidence
192	GPT-4.1 nano	32.4%	estimated ± 1.8 pp, low confidence
193	Nova Pro	31.3%	estimated ± 1.8 pp, low confidence
194	Ultravox v0.6 Llama 3.3 70B	31.2%	estimated ± 1.8 pp, low confidence
195	Claude 3 Opus	30.5%	estimated ± 1.8 pp, low confidence
196	Mistral Large 2	30.2%	estimated ± 1.8 pp, low confidence
197	Nemotron 3 Nano Omni 30B A3B	28.9%	estimated ± 1.8 pp, low confidence
198	Gemma 4 E2B	26.0%	estimated ± 1.8 pp, low confidence
199	Gemma 3 27B	25.6%	estimated ± 1.8 pp, low confidence
200	GPT-4o mini	25.5%	estimated ± 1.8 pp, low confidence
201	Exaone 4.0 1.2B	25.3%	estimated ± 1.8 pp, low confidence
202	Qwen2.5 Coder 32B Instruct	24.8%	estimated ± 1.8 pp, low confidence
203	GPT-5.4 mini	22.8%	estimated ± 1.3 pp, low confidence
204	Claude Sonnet 4.5 Thinking	22.7%	estimated ± 1.3 pp, low confidence
205	Claude 3 Haiku	21.7%	estimated ± 1.8 pp, low confidence
206	Phi-4 Multimodal Instruct	17.6%	estimated ± 1.8 pp, low confidence
207	LFM2.5-VL-1.6B-Extract	15.9%	estimated ± 1.8 pp, low confidence
208	Gemini 1.0 Pro	15.1%	estimated ± 1.8 pp, low confidence
209	Granite-4.0-H-1B	14.3%	estimated ± 1.8 pp, low confidence
210	Granite-4.0-350M	14.1%	estimated ± 1.8 pp, low confidence
211	Granite-4.0-H-350M	13.9%	estimated ± 1.8 pp, low confidence
212	GPT-5.4 nano	11.5%	estimated ± 1.3 pp, low confidence

LABBench2 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General