Agentic · tools

AA Agentic Index leaderboard

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

As of 2026-10-07, the highest measured score on AA Agentic Index is 58.0% by Claude Fable 5.1. 151 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5.5	61.3%	estimated ± 2.8 pp, medium confidence
2	Claude Sonnet 5.5	60.3%	estimated ± 2.8 pp, medium confidence
3	Claude Fable 5.1	58.0%	measured
4	GLM-5.3-Flash	56.2%	estimated ± 3.3 pp, medium confidence
5	Claude Opus 5	56.2%	measured
6	MiMo-V2.6-Pro	56.0%	estimated ± 1.7 pp, high confidence
7	Muse Spark 1.3	55.7%	measured
8	Grok 4.7	55.1%	estimated ± 2.8 pp, high confidence
9	GLM-5.3	53.4%	measured
10	Grok 4.6	53.4%	measured
11	Fugu Cyber	53.1%	estimated ± 5.8 pp, low confidence
12	MiMo-V2.6-Flash	52.7%	estimated ± 1.7 pp, high confidence
13	Atria Dawn Preview	52.5%	estimated ± 5.8 pp, low confidence
14	Ling 3.1 Flash	52.0%	estimated ± 1.7 pp, high confidence
15	Gemini 3.8 Flash Cyber	52.0%	estimated ± 5.8 pp, low confidence
16	GPT-6 Astra	51.5%	measured
17	Qwen3.8-Flash-Next	51.4%	estimated ± 2.8 pp, high confidence
18	Gemini 4 Argon	51.1%	estimated ± 2.8 pp, high confidence
19	DeepSeek V4.1 Flash	51.0%	estimated ± 1.7 pp, high confidence
20	Claude Fable 5	51.0%	measured
21	Kimi K3	50.6%	measured
22	Hy4 preview	50.6%	estimated ± 3.3 pp, medium confidence
23	GPT-5.6 Sol	50.5%	measured
24	Step 5 Preview	49.8%	estimated ± 1.7 pp, high confidence
25	Qwen3.8 Max Preview	49.6%	measured
26	DeepSeek V4 Pro 0813	49.6%	measured
27	GPT-5.5 Pro	49.1%	estimated ± 8.2 pp, medium confidence
28	GPT-6.1 Sol	48.8%	estimated ± 2.8 pp, high confidence
29	Qwen3.8 Max	48.5%	estimated ± 3.3 pp, medium confidence
30	Claude Mythos 5	48.5%	estimated ± 5.8 pp, medium confidence
31	Gemini 3.5 Flash Cyber	47.7%	estimated ± 5.8 pp, medium confidence
32	GPT-5.4 Pro	47.6%	estimated ± 8.2 pp, medium confidence
33	Claude Mythos Preview	47.5%	estimated ± 5.8 pp, medium confidence
34	Ornith-1.5-397B	46.8%	estimated ± 3.3 pp, medium confidence
35	Qwen3.8-27B	46.5%	measured
36	GPT-6 Sol	45.6%	estimated ± 2.8 pp, high confidence
37	Claude Sonnet 5	44.3%	measured
38	Muse Spark 1.2	44.0%	measured
39	GPT-5.6 Terra	43.7%	measured
40	GPT-5.6 Luna	42.7%	measured
41	Claude Opus 4.8	42.6%	measured
42	Grok 4.5	42.1%	measured
43	GPT-6 Luna	42.1%	estimated ± 2.8 pp, high confidence
44	DeepSeek V4 Flash 0731	41.7%	measured
45	Mistral Large 4	41.4%	estimated ± 2.8 pp, high confidence
46	Gemini 3.8 Flash	41.1%	measured
47	Holo3-35B-A3B	39.8%	estimated ± 8.0 pp, medium confidence
48	Claude Opus 4.7 (Adaptive)	39.5%	measured
49	GLM-5.2	39.4%	measured
50	GPT-5.5	37.3%	measured
51	Claude Haiku 5.5	37.1%	estimated ± 6.7 pp, medium confidence
52	Gemini 3.7 Flash	36.4%	measured
53	GPT-5.4	33.2%	estimated ± 1.7 pp, high confidence
54	Claude Opus 4.6 (Adaptive)	32.9%	estimated ± 11.8 pp, low confidence
55	Quasar 438B	32.7%	measured
56	Holo3-122B-A10B	31.5%	estimated ± 8.0 pp, medium confidence
57	Beam	31.1%	estimated ± 8.2 pp, medium confidence
58	MiniMax M3	30.8%	measured
59	Gemini 3.6 Flash	30.2%	measured
60	Apodex 1.1	29.4%	estimated ± 2.8 pp, high confidence
61	Apodex 1.1 Mini	29.4%	estimated ± 2.8 pp, high confidence
62	Agents-A1	29.2%	estimated ± 8.2 pp, medium confidence
63	Claude Opus 4.7	28.7%	estimated ± 6.4 pp, medium confidence
64	UI-Mate-27B	28.3%	estimated ± 8.0 pp, medium confidence
65	K-EXAONE 2.0	27.7%	estimated ± 5.1 pp, low confidence
66	Ling 3.0 Flash VL	27.6%	estimated ± 2.8 pp, high confidence
67	Muse Spark 1.1	27.5%	measured
68	Ornith-1.0-397B	27.4%	estimated ± 5.1 pp, low confidence
69	Gemini 3.5 Flash	27.3%	measured
70	dots3-note Preview	26.5%	estimated ± 3.3 pp, medium confidence
71	Gemini 3 Pro	26.2%	estimated ± 6.4 pp, medium confidence
72	Hy3	25.6%	measured
73	Hy3 Preview	25.6%	measured
74	GLM-5.1	25.2%	measured
75	Inkling-Small	25.0%	measured
76	Qwen 3.6 Max (preview)	24.8%	estimated ± 12.3 pp, low confidence
77	Solar Pro 4	24.7%	estimated ± 2.8 pp, high confidence
78	Inkling	24.3%	measured
79	Qwen3.7 Max	23.9%	measured
80	Grok 4.1 Fast (Reasoning)	23.9%	estimated ± 12.3 pp, low confidence
81	Claude Opus 4.6	23.7%	estimated ± 5.1 pp, medium confidence
82	Ornith-1.0-35B	23.4%	estimated ± 5.1 pp, medium confidence
83	MiMo-V2.5-Pro	22.7%	measured
84	Kimi K2.7 Code	22.5%	measured
85	Claude Opus 4.5 Thinking	22.5%	estimated ± 12.3 pp, low confidence
86	Claude Sonnet 4.6	22.4%	estimated ± 5.1 pp, medium confidence
87	Kimi K2.6	22.1%	measured
88	Agents-A1-4B	22.0%	estimated ± 8.2 pp, medium confidence
89	LFM2.5-8B-A1B	22.0%	estimated ± 12.3 pp, low confidence
90	Nemotron 3 Ultra	21.7%	measured
91	GPT-5 (medium)	21.4%	estimated ± 12.3 pp, low confidence
92	GPT-5.3 Codex	21.1%	estimated ± 6.4 pp, medium confidence
93	Ling 3.0 Flash	21.0%	measured
94	Ling 3.0 Flash FP8	21.0%	measured
95	LLaDA2.2-flash	20.6%	estimated ± 5.1 pp, medium confidence
96	Qwen3.5 397B (Reasoning)	20.4%	estimated ± 12.3 pp, low confidence
97	Ornith-1.0-9B	20.1%	estimated ± 5.1 pp, medium confidence
98	GPT-5.1-Codex-Max	20.1%	estimated ± 12.3 pp, low confidence
99	Qwen3.6-27B	20.1%	measured
100	GLM-4.7	19.8%	estimated ± 2.8 pp, high confidence
101	Step 3.7 Flash	19.8%	estimated ± 2.8 pp, high confidence
102	MiMo-V2.5	19.7%	estimated ± 5.1 pp, medium confidence
103	Qwen3.7 Plus	19.7%	measured
104	GPT-5.4 mini	19.7%	measured
105	o3	19.2%	estimated ± 12.3 pp, low confidence
106	Ternary Bonsai 2 27B	19.1%	estimated ± 12.3 pp, low confidence
107	Laguna S 2.1	18.8%	estimated ± 3.3 pp, low confidence
108	Qwen3.6 Plus	18.7%	estimated ± 2.8 pp, high confidence
109	Claude Opus 4.5	18.5%	estimated ± 5.1 pp, medium confidence
110	GLM-4.6	17.8%	estimated ± 12.3 pp, low confidence
111	GPT-5.4 nano	17.7%	measured
112	MiMo-V2-Pro	17.7%	estimated ± 5.1 pp, medium confidence
113	GLM-5	17.7%	estimated ± 5.1 pp, medium confidence
114	Ornith-1.5-35B-A3B	17.5%	estimated ± 3.3 pp, low confidence
115	LLaDA2.2-mini	17.5%	estimated ± 5.1 pp, medium confidence
116	Grok Code Fast 1	17.4%	estimated ± 12.3 pp, low confidence
117	Qwen3.5 397B	17.3%	estimated ± 5.1 pp, medium confidence
118	Grok 4.3	17.2%	measured
119	A.X K2	17.2%	estimated ± 2.8 pp, high confidence
120	GPT-5.2-Codex	17.0%	estimated ± 6.4 pp, medium confidence
121	GLM-5-Turbo	16.9%	estimated ± 5.1 pp, medium confidence
122	MiniMax M2.7	16.8%	measured
123	UI-Mate-9B	16.7%	estimated ± 8.0 pp, medium confidence
124	GLM-5V-Turbo	16.1%	estimated ± 5.1 pp, medium confidence
125	Gemini 3.5 Flash-Lite	15.9%	measured
126	Claude 4.1 Opus Thinking	15.8%	estimated ± 12.3 pp, low confidence
127	Muse Spark	15.8%	estimated ± 1.7 pp, high confidence
128	GPT-5.1-Codex	15.7%	estimated ± 6.4 pp, medium confidence
129	Grok 4.1	15.5%	estimated ± 6.8 pp, medium confidence
130	Grok Build 0.1	15.4%	estimated ± 6.4 pp, medium confidence
131	Claude Sonnet 4.5	15.1%	estimated ± 6.4 pp, medium confidence
132	GPT-5 (high)	15.0%	estimated ± 2.8 pp, high confidence
133	Qwen3.6-35B-A3B	15.0%	measured
134	Grok 4.1 Fast	14.4%	estimated ± 6.4 pp, medium confidence
135	Gemini 3 Flash	14.2%	estimated ± 5.1 pp, medium confidence
136	GPT-5.2	14.0%	estimated ± 6.4 pp, medium confidence
137	Grok 4 Fast (Reasoning)	13.8%	estimated ± 12.3 pp, low confidence
138	MiMo-V2-Omni	12.7%	estimated ± 5.1 pp, medium confidence
139	o1	12.6%	estimated ± 12.3 pp, low confidence
140	Qwen3 Max	12.6%	estimated ± 6.4 pp, medium confidence
141	LongCat-Flash-Lite-Sparse	12.2%	estimated ± 8.2 pp, medium confidence
142	Kimi K2	12.0%	estimated ± 12.3 pp, low confidence
143	Grok 4	11.9%	estimated ± 6.4 pp, medium confidence
144	Kimi K2.5	11.6%	estimated ± 2.8 pp, high confidence
145	Kimi K2.5 (Reasoning)	11.6%	estimated ± 2.8 pp, high confidence
146	GPT-5.1	11.0%	estimated ± 2.8 pp, high confidence
147	DeepSeek V3.2	11.0%	estimated ± 5.1 pp, medium confidence
148	Claude 4 Sonnet	10.7%	estimated ± 6.4 pp, medium confidence
149	Qwen3.5-27B	10.6%	estimated ± 6.4 pp, medium confidence
150	Muse Glimmer 30B	10.5%	measured
151	Gemini 3.1 Pro	10.3%	measured
152	Gemini 3.1 Flash-Lite	10.2%	estimated ± 6.4 pp, medium confidence
153	Grok 4.20	10.2%	estimated ± 6.4 pp, medium confidence
154	Qwen3.5-122B-A10B	9.6%	measured
155	Mistral Medium 3.5 128B	9.4%	measured
156	Ornith-1.5-9B	7.7%	estimated ± 3.3 pp, low confidence
157	Sarvam 105B	6.8%	estimated ± 12.3 pp, low confidence
158	Qwen3.5-35B-A3B	6.8%	estimated ± 6.4 pp, low confidence
159	Gemma 4 31B	6.7%	measured
160	GLM-4.5-Air	6.7%	estimated ± 12.3 pp, low confidence
161	MiniCPM5-2B	6.5%	estimated ± 2.8 pp, high confidence
162	GPT-OSS 120B	6.2%	measured
163	Nemotron 3.5 Lightning 30B A3B NVFP4	6.1%	measured
164	GPT-4.1	5.7%	estimated ± 6.4 pp, low confidence
165	MiMo-V2-Flash	4.4%	estimated ± 2.8 pp, high confidence
166	Nemotron 3 Super 100B	4.1%	measured
167	Granite 4.2 8B	3.7%	measured
168	Command A+	3.6%	measured
169	Gemini 2.5 Pro	3.6%	measured
170	Granite 4.2 30B	3.3%	estimated ± 2.8 pp, high confidence
171	DeepSeek V3.1 (Reasoning)	3.3%	estimated ± 12.3 pp, low confidence
172	Gemma 4 26B A4B	3.2%	estimated ± 2.8 pp, high confidence
173	Ling 3.0 Tiny	3.1%	estimated ± 2.8 pp, high confidence
174	DeepSeek-R1	3.0%	estimated ± 12.3 pp, low confidence
175	DeepSeek V3 0324	2.7%	estimated ± 2.8 pp, high confidence
176	Gemma 4 12B	2.7%	estimated ± 2.8 pp, high confidence
177	Gemma 4 E2B	2.7%	estimated ± 2.8 pp, high confidence
178	Gemma 4 E4B	2.7%	estimated ± 2.8 pp, high confidence
179	GPT-4.1 mini	2.7%	estimated ± 2.8 pp, high confidence
180	GPT-4.1 nano	2.7%	estimated ± 2.8 pp, high confidence
181	GPT-4o	2.7%	estimated ± 2.8 pp, high confidence
182	GPT-4o mini	2.7%	estimated ± 2.8 pp, high confidence
183	Granite 4.2 3B	2.7%	estimated ± 2.8 pp, high confidence
184	K-Exaone	2.7%	estimated ± 2.8 pp, high confidence
185	LFM2.5-2.6B	2.7%	estimated ± 2.8 pp, high confidence
186	Ling 2.6 Flash	2.7%	estimated ± 2.8 pp, high confidence
187	Mercury 2.5	2.7%	estimated ± 2.8 pp, high confidence
188	Nemotron 3 Nano Omni 30B A3B	2.7%	estimated ± 2.8 pp, high confidence
189	North Mini Code	2.7%	estimated ± 2.8 pp, high confidence
190	Solar Pro 3	2.7%	estimated ± 2.8 pp, high confidence
191	Ultravox v0.6 Llama 3.3 70B	2.7%	estimated ± 2.8 pp, high confidence
192	Mistral Large 3	2.4%	measured
193	DeepSeek V3.1	2.4%	estimated ± 12.3 pp, low confidence
194	Sarvam 30B	2.3%	estimated ± 12.3 pp, low confidence
195	Mistral Small 4	1.4%	measured
196	Mistral Small 4 (Reasoning)	1.4%	measured
197	GPT-OSS 20B	1.4%	measured
198	Solar Pro 2	1.3%	estimated ± 12.3 pp, low confidence
199	Trinity-Large-Preview	1.2%	measured
200	Trinity-Large-Thinking	1.2%	measured
201	Nemotron 3 Nano 30B	1.0%	measured
202	Mistral Large 2	0.9%	estimated ± 12.3 pp, low confidence
203	DeepSeek V3	0.8%	measured
204	Celeris-1	0.7%	measured
205	Llama 4 Maverick	0.6%	measured
206	Llama 4 Scout	0.5%	measured
207	Gemma 3 27B	0.1%	measured
208	o3-mini	0.1%	estimated ± 12.3 pp, low confidence
209	Claude 3 Haiku	0.0%	estimated ± 12.3 pp, low confidence
210	Exaone 4.0 1.2B	0.0%	estimated ± 12.3 pp, low confidence
211	Exaone 4.0 32B	0.0%	estimated ± 12.3 pp, low confidence
212	Gemini 2.5 Flash	0.0%	estimated ± 12.3 pp, low confidence
213	Granite-4.0-350M	0.0%	estimated ± 12.3 pp, low confidence
214	Granite-4.0-H-1B	0.0%	estimated ± 12.3 pp, low confidence
215	Granite-4.0-H-350M	0.0%	estimated ± 12.3 pp, low confidence
216	LFM2.5-VL-1.6B-Extract	0.0%	estimated ± 12.3 pp, low confidence
217	Llama 3.1 405B	0.0%	estimated ± 12.3 pp, low confidence
218	Mistral Medium 3	0.0%	estimated ± 12.3 pp, low confidence
219	Nemotron Ultra 253B	0.0%	estimated ± 12.3 pp, low confidence
220	Nova Pro	0.0%	estimated ± 12.3 pp, low confidence
221	Phi-4	0.0%	estimated ± 12.3 pp, low confidence
222	Qwen3-Omni-30B-A3B-Instruct	0.0%	estimated ± 12.3 pp, low confidence
223	Qwen3-Omni-30B-A3B-Thinking	0.0%	estimated ± 12.3 pp, low confidence

AA Agentic Index leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General