Agentic · tools

APEX-Agents-AA leaderboard

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

As of 2026-10-07, the highest measured score on APEX-Agents-AA is 47.1% by Gemini 3.5 Flash. 189 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Sonnet 5.5	48.1%	estimated ± 9.1 pp, low confidence
2	Gemini 3.5 Flash	47.1%	measured
3	Claude Fable 5	45.0%	estimated ± 3.5 pp, low confidence
4	Qwen3.8 Max Preview	42.4%	measured
5	MiMo-V2.6-Pro	41.7%	estimated ± 3.6 pp, low confidence
6	Hy4 preview	41.7%	estimated ± 3.6 pp, low confidence
7	MiMo-V2.6-Flash	41.5%	estimated ± 3.6 pp, low confidence
8	Kimi K3	41.3%	measured
9	GPT-5.5 Pro	41.2%	estimated ± 4.6 pp, high confidence
10	GPT-5.4 Pro	39.9%	estimated ± 4.6 pp, high confidence
11	GPT-6.1 Sol	39.6%	estimated ± 9.1 pp, medium confidence
12	Ling 3.1 Flash	39.5%	estimated ± 6.0 pp, low confidence
13	GLM-5.3-Flash	39.4%	estimated ± 3.2 pp, medium confidence
14	DeepSeek V4.1 Flash	39.2%	estimated ± 3.2 pp, medium confidence
15	GPT-6 Astra	38.9%	estimated ± 2.9 pp, low confidence
16	GPT-5.6 Terra	38.9%	measured
17	Claude Opus 5	38.9%	estimated ± 2.9 pp, low confidence
18	Gemini 4 Argon	38.9%	estimated ± 2.9 pp, low confidence
19	Muse Spark 1.3	38.8%	estimated ± 2.9 pp, low confidence
20	GPT-5.6 Sol	38.7%	estimated ± 2.9 pp, low confidence
21	GPT-6 Sol	38.7%	estimated ± 2.9 pp, low confidence
22	Fugu Cyber	38.7%	estimated ± 6.0 pp, low confidence
23	Gemini 3.8 Flash	38.7%	estimated ± 2.9 pp, low confidence
24	Claude Opus 5.5	38.4%	estimated ± 2.9 pp, medium confidence
25	Gemini 3.7 Flash	38.4%	estimated ± 2.9 pp, medium confidence
26	GLM-5.3	38.4%	estimated ± 3.2 pp, medium confidence
27	Atria Dawn Preview	38.3%	estimated ± 3.6 pp, medium confidence
28	Claude Fable 5.1	38.2%	estimated ± 2.9 pp, medium confidence
29	Gemini 3.8 Flash Cyber	38.1%	estimated ± 6.0 pp, low confidence
30	Claude Sonnet 5	38.0%	estimated ± 3.5 pp, low confidence
31	Step 5 Preview	38.0%	measured
32	Claude Mythos 5	37.8%	estimated ± 4.6 pp, high confidence
33	GPT-5.5	37.7%	measured
34	Grok 4.6	36.9%	estimated ± 3.5 pp, low confidence
35	Claude Opus 4.8	36.5%	estimated ± 2.9 pp, medium confidence
36	Muse Spark 1.2	36.4%	estimated ± 9.1 pp, medium confidence
37	Qwen3.8-Flash-Next	36.3%	estimated ± 2.9 pp, medium confidence
38	Qwen3.8 Max	36.3%	estimated ± 2.9 pp, medium confidence
39	Claude Opus 4.7 (Adaptive)	36.1%	estimated ± 2.9 pp, medium confidence
40	Gemini 3.5 Flash Cyber	35.9%	estimated ± 6.0 pp, low confidence
41	GPT-5.6 Luna	35.8%	measured
42	Claude Mythos Preview	35.8%	estimated ± 6.0 pp, low confidence
43	Ornith-1.5-397B	35.6%	estimated ± 4.6 pp, high confidence
44	Muse Spark 1.1	35.2%	estimated ± 2.9 pp, medium confidence
45	GPT-6 Luna	35.1%	estimated ± 9.1 pp, medium confidence
46	Claude Opus 4.7	35.1%	estimated ± 2.9 pp, medium confidence
47	Qwen3.7 Max	34.8%	estimated ± 5.7 pp, medium confidence
48	Mistral Large 4	34.6%	estimated ± 9.1 pp, medium confidence
49	Kimi K2.7 Code	34.6%	estimated ± 5.7 pp, medium confidence
50	Muse Glimmer 30B	34.3%	estimated ± 5.7 pp, medium confidence
51	Claude Opus 4.6	33.8%	estimated ± 3.5 pp, low confidence
52	GLM-5.2	33.7%	measured
53	Grok 4.7	33.7%	estimated ± 3.2 pp, low confidence
54	Grok 4.5	33.5%	estimated ± 9.1 pp, medium confidence
55	GPT-5.4	33.3%	measured
56	Claude Opus 4.6 (Adaptive)	33.0%	measured
57	Claude Sonnet 4.6	32.4%	estimated ± 2.9 pp, medium confidence
58	Gemini 3.1 Pro	32.0%	measured
59	GPT-5.2	31.9%	estimated ± 3.6 pp, medium confidence
60	MiMo-V2.5	31.8%	estimated ± 10.0 pp, low confidence
61	GPT-5.3 Codex	31.6%	estimated ± 3.6 pp, medium confidence
62	Qwen3.8-27B	31.5%	estimated ± 3.6 pp, medium confidence
63	Apodex 1.1	31.2%	measured
64	Apodex 1.1 Mini	31.2%	measured
65	Claude Opus 4.5	30.9%	estimated ± 3.6 pp, medium confidence
66	dots3-note Preview	30.8%	estimated ± 4.6 pp, high confidence
67	Gemini 3.6 Flash	30.2%	estimated ± 9.1 pp, medium confidence
68	MiMo-V2-Pro	29.2%	estimated ± 10.0 pp, low confidence
69	Kimi K2.6	28.5%	measured
70	Claude Sonnet 4.5	28.4%	estimated ± 3.6 pp, medium confidence
71	Qwen3.6-35B-A3B	28.3%	estimated ± 5.7 pp, medium confidence
72	GPT-5.4 mini	28.2%	measured
73	MiniMax M3	28.2%	estimated ± 2.9 pp, medium confidence
74	Hy3 Preview	27.9%	estimated ± 9.1 pp, medium confidence
75	GPT-5.1-Codex	27.4%	estimated ± 3.6 pp, medium confidence
76	GPT-5.2-Codex	27.3%	estimated ± 3.6 pp, medium confidence
77	Quasar 438B	26.9%	estimated ± 9.1 pp, medium confidence
78	Ling 3.0 Flash VL	26.2%	estimated ± 9.1 pp, medium confidence
79	Grok 4.1	26.2%	estimated ± 10.0 pp, low confidence
80	Solar Open 2	26.1%	estimated ± 5.7 pp, medium confidence
81	GLM-5-Turbo	25.3%	estimated ± 11.4 pp, low confidence
82	GLM-5V-Turbo	25.3%	estimated ± 11.4 pp, low confidence
83	Grok 4.1 Fast (Reasoning)	25.3%	estimated ± 11.4 pp, low confidence
84	Qwen 3.6 Max (preview)	25.3%	estimated ± 11.4 pp, low confidence
85	MiMo-V2-Omni	25.3%	estimated ± 11.4 pp, low confidence
86	Claude Opus 4.5 Thinking	25.3%	estimated ± 11.4 pp, low confidence
87	LFM2.5-8B-A1B	25.2%	estimated ± 11.4 pp, low confidence
88	GPT-5.4 nano	24.9%	measured
89	Claude 4.1 Opus	24.6%	estimated ± 3.6 pp, medium confidence
90	GPT-5 (medium)	24.5%	estimated ± 11.4 pp, low confidence
91	DeepSeek V4 Pro 0813	24.3%	measured
92	Inkling-Small	23.2%	estimated ± 4.6 pp, high confidence
93	Beam	23.2%	estimated ± 4.6 pp, high confidence
94	Hy3	23.0%	estimated ± 9.1 pp, medium confidence
95	Inkling	22.9%	estimated ± 4.6 pp, high confidence
96	Qwen3.7 Plus	22.4%	measured
97	Qwen3.5 Plus	22.0%	estimated ± 3.6 pp, medium confidence
98	Claude 4 Sonnet	21.9%	estimated ± 3.6 pp, medium confidence
99	Qwen3.6 Plus	21.4%	estimated ± 5.7 pp, medium confidence
100	Agents-A1	21.1%	estimated ± 4.6 pp, high confidence
101	Qwen3.6-27B	20.5%	estimated ± 9.1 pp, medium confidence
102	Gemini 3.5 Flash-Lite	20.4%	estimated ± 9.1 pp, medium confidence
103	LLaDA2.2-flash	20.4%	estimated ± 5.7 pp, medium confidence
104	Claude Haiku 4.5	19.9%	estimated ± 3.6 pp, medium confidence
105	A.X K2	19.7%	estimated ± 9.1 pp, medium confidence
106	Ling 3.0 Flash FP8	19.2%	estimated ± 9.1 pp, medium confidence
107	DeepSeek V4 Flash 0731	18.8%	estimated ± 4.6 pp, high confidence
108	Grok Build 0.1	18.5%	estimated ± 11.4 pp, low confidence
109	Ling 3.0 Flash	17.9%	estimated ± 4.6 pp, high confidence
110	Grok 4.3	17.0%	measured
111	Gemini 3 Flash	15.5%	estimated ± 3.6 pp, medium confidence
112	Gemini 3 Pro	15.5%	estimated ± 3.6 pp, medium confidence
113	GPT-5.1	15.4%	estimated ± 9.1 pp, medium confidence
114	Step 3.7 Flash	14.8%	measured
115	GLM-5.1	14.5%	estimated ± 4.6 pp, high confidence
116	GLM-5	14.5%	measured
117	Ornith-1.5-35B-A3B	14.2%	estimated ± 4.6 pp, high confidence
118	Muse Spark	14.0%	estimated ± 6.0 pp, low confidence
119	Agents-A1-4B	13.7%	estimated ± 4.6 pp, high confidence
120	Mistral Medium 3.5 128B	13.2%	estimated ± 9.1 pp, medium confidence
121	GPT-5 (high)	12.3%	estimated ± 3.6 pp, low confidence
122	Qwen3.5-122B-A10B	11.9%	estimated ± 4.6 pp, high confidence
123	Kimi K2.5	11.5%	measured
124	Kimi K2.5 (Reasoning)	11.5%	measured
125	MiniCPM5-2B	11.5%	estimated ± 9.1 pp, medium confidence
126	Qwen3.5 397B	10.9%	estimated ± 4.6 pp, high confidence
127	Gemini 3.1 Flash-Lite	10.6%	estimated ± 11.4 pp, low confidence
128	MiniMax M2.7	10.6%	measured
129	Grok 4.20	10.5%	estimated ± 11.4 pp, low confidence
130	Qwen3.5-27B	10.5%	estimated ± 4.6 pp, high confidence
131	Qwen3.5-35B-A3B	10.5%	estimated ± 4.6 pp, high confidence
132	MiMo-V2-Flash	9.0%	estimated ± 9.1 pp, medium confidence
133	Ornith-1.5-9B	8.8%	estimated ± 4.6 pp, medium confidence
134	Gemma 4 31B	8.6%	estimated ± 9.1 pp, medium confidence
135	GLM-4.7	7.6%	estimated ± 4.6 pp, medium confidence
136	Granite 4.2 30B	7.2%	estimated ± 9.1 pp, medium confidence
137	Solar Pro 4	7.1%	estimated ± 4.6 pp, medium confidence
138	LongCat-Flash-Lite-Sparse	7.0%	estimated ± 4.6 pp, medium confidence
139	Gemma 4 26B A4B	6.9%	estimated ± 9.1 pp, medium confidence
140	Ling 3.0 Tiny	6.8%	estimated ± 9.1 pp, medium confidence
141	Nemotron 3 Ultra	6.5%	estimated ± 4.6 pp, medium confidence
142	Nemotron 3.5 Lightning 30B A3B NVFP4	6.0%	estimated ± 4.6 pp, medium confidence
143	Command A+	5.3%	estimated ± 9.1 pp, medium confidence
144	Qwen3.5 397B (Reasoning)	4.9%	estimated ± 11.4 pp, low confidence
145	Celeris-1	4.7%	estimated ± 9.1 pp, medium confidence
146	DeepSeek V3	4.7%	estimated ± 9.1 pp, medium confidence
147	DeepSeek V3 0324	4.7%	estimated ± 9.1 pp, medium confidence
148	Gemini 2.5 Pro	4.7%	estimated ± 9.1 pp, medium confidence
149	Gemma 3 27B	4.7%	estimated ± 9.1 pp, medium confidence
150	Gemma 4 12B	4.7%	estimated ± 9.1 pp, medium confidence
151	Gemma 4 E2B	4.7%	estimated ± 9.1 pp, medium confidence
152	Gemma 4 E4B	4.7%	estimated ± 9.1 pp, medium confidence
153	GPT-4.1 mini	4.7%	estimated ± 9.1 pp, medium confidence
154	GPT-4.1 nano	4.7%	estimated ± 9.1 pp, medium confidence
155	GPT-4o	4.7%	estimated ± 9.1 pp, medium confidence
156	GPT-4o mini	4.7%	estimated ± 9.1 pp, medium confidence
157	Granite 4.2 3B	4.7%	estimated ± 9.1 pp, medium confidence
158	Granite 4.2 8B	4.7%	estimated ± 9.1 pp, medium confidence
159	K-Exaone	4.7%	estimated ± 9.1 pp, medium confidence
160	LFM2.5-2.6B	4.7%	estimated ± 9.1 pp, medium confidence
161	Ling 2.6 Flash	4.7%	estimated ± 9.1 pp, medium confidence
162	Llama 4 Maverick	4.7%	estimated ± 9.1 pp, medium confidence
163	Llama 4 Scout	4.7%	estimated ± 9.1 pp, medium confidence
164	Mercury 2.5	4.7%	estimated ± 9.1 pp, medium confidence
165	Mistral Large 3	4.7%	estimated ± 9.1 pp, medium confidence
166	Mistral Small 4	4.7%	estimated ± 9.1 pp, medium confidence
167	Mistral Small 4 (Reasoning)	4.7%	estimated ± 9.1 pp, medium confidence
168	Nemotron 3 Nano 30B	4.7%	estimated ± 9.1 pp, medium confidence
169	Nemotron 3 Nano Omni 30B A3B	4.7%	estimated ± 9.1 pp, medium confidence
170	North Mini Code	4.7%	estimated ± 9.1 pp, medium confidence
171	Solar Pro 3	4.7%	estimated ± 9.1 pp, medium confidence
172	Trinity-Large-Preview	4.7%	estimated ± 9.1 pp, medium confidence
173	Trinity-Large-Thinking	4.7%	estimated ± 9.1 pp, medium confidence
174	Ultravox v0.6 Llama 3.3 70B	4.7%	estimated ± 9.1 pp, medium confidence
175	GPT-OSS 120B	3.1%	measured
176	MiMo-V2.5-Pro	2.4%	measured
177	GPT-5.1-Codex-Max	2.3%	estimated ± 11.4 pp, low confidence
178	Nemotron 3 Super 100B	1.8%	measured
179	o3	1.7%	estimated ± 11.4 pp, low confidence
180	Ternary Bonsai 2 27B	1.6%	estimated ± 11.4 pp, low confidence
181	DeepSeek V3.2	1.6%	estimated ± 11.4 pp, low confidence
182	Claude 3 Haiku	1.6%	estimated ± 11.4 pp, low confidence
183	Claude 4.1 Opus Thinking	1.6%	estimated ± 11.4 pp, low confidence
184	DeepSeek-R1	1.6%	estimated ± 11.4 pp, low confidence
185	DeepSeek V3.1	1.6%	estimated ± 11.4 pp, low confidence
186	DeepSeek V3.1 (Reasoning)	1.6%	estimated ± 11.4 pp, low confidence
187	Exaone 4.0 1.2B	1.6%	estimated ± 11.4 pp, low confidence
188	Exaone 4.0 32B	1.6%	estimated ± 11.4 pp, low confidence
189	Gemini 2.5 Flash	1.6%	estimated ± 11.4 pp, low confidence
190	GLM-4.5-Air	1.6%	estimated ± 11.4 pp, low confidence
191	GLM-4.6	1.6%	estimated ± 11.4 pp, low confidence
192	GPT-4.1	1.6%	estimated ± 11.4 pp, low confidence
193	Granite-4.0-350M	1.6%	estimated ± 11.4 pp, low confidence
194	Granite-4.0-H-1B	1.6%	estimated ± 11.4 pp, low confidence
195	Granite-4.0-H-350M	1.6%	estimated ± 11.4 pp, low confidence
196	Grok 4	1.6%	estimated ± 11.4 pp, low confidence
197	Grok 4.1 Fast	1.6%	estimated ± 11.4 pp, low confidence
198	Grok 4 Fast (Reasoning)	1.6%	estimated ± 11.4 pp, low confidence
199	Grok Code Fast 1	1.6%	estimated ± 11.4 pp, low confidence
200	Kimi K2	1.6%	estimated ± 11.4 pp, low confidence
201	LFM2.5-VL-1.6B-Extract	1.6%	estimated ± 11.4 pp, low confidence
202	LLaDA2.2-mini	1.6%	estimated ± 11.4 pp, low confidence
203	Llama 3.1 405B	1.6%	estimated ± 11.4 pp, low confidence
204	Mistral Large 2	1.6%	estimated ± 11.4 pp, low confidence
205	Mistral Medium 3	1.6%	estimated ± 11.4 pp, low confidence
206	Nemotron Ultra 253B	1.6%	estimated ± 11.4 pp, low confidence
207	Nova Pro	1.6%	estimated ± 11.4 pp, low confidence
208	o1	1.6%	estimated ± 11.4 pp, low confidence
209	o3-mini	1.6%	estimated ± 11.4 pp, low confidence
210	Phi-4	1.6%	estimated ± 11.4 pp, low confidence
211	Qwen3 Max	1.6%	estimated ± 11.4 pp, low confidence
212	Qwen3-Omni-30B-A3B-Instruct	1.6%	estimated ± 11.4 pp, low confidence
213	Qwen3-Omni-30B-A3B-Thinking	1.6%	estimated ± 11.4 pp, low confidence
214	Sarvam 105B	1.6%	estimated ± 11.4 pp, low confidence
215	Sarvam 30B	1.6%	estimated ± 11.4 pp, low confidence
216	Solar Pro 2	1.6%	estimated ± 11.4 pp, low confidence
217	GPT-OSS 20B	0.7%	measured

APEX-Agents-AA leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General