Benchmarks · agentic

APEX-Agents-AA

APEX-Agents-AA: long-horizon professional-services agent tasks; pass rate.

21 models · 21 results · data as of 2026-10-29 · artificialanalysis.ai ↗

Breakdown

darker is better within each column · click a column to sort

Model
Gemini 3.5 Flash high	1,148,898,405	9,203,624	13,058,455
Kimi K3 max	279,424,618	2,689,185	6,196,153
GPT-5.6 Terra max	433,643,836	2,907,783	4,911,382
GPT-5.5 xhigh	252,687,804	2,267,228	2,495,886
GPT-5.6 Luna max	503,230,610	3,004,789	4,463,083
GLM 5.2 max	416,930,057	4,394,399	26,006,369
GPT-5.4 xhigh	351,072,983	2,845,641	12,614,930
Kimi K2.6	762,518,111	6,045,625	7,736,074
GPT-5.4 Mini xhigh	382,648,221	2,675,988	34,091,237
GPT-5.4 Nano xhigh	1,360,318,828	4,418,068	107,247,662
DeepSeek V4 Pro 0424 max	605,993,824	5,211,273	4,596,648
Qwen 3.7 Plus	272,042,704	3,690,984	6,903,348
Grok 4.3 high	284,663,761	1,288,051	3,157,470
Qwen 3.5 397B-A17B	317,868,020	4,563,624	533,914
GLM 5	486,704,512	3,033,970	3,091,817
Kimi K2.5	239,659,057	2,764,061	1,695,232
MiniMax M2.7	416,213,643	3,873,551	4,301,920
gpt-oss-120b high	546,855,639	5,149,676	6,377,259
MiMo V2.5 Pro	2,247,579,005	3,970,256	24,403,071
gpt-oss-20b high	55,761,208	788,351	4,073,223

All results

21 rows

#	Model	Effort	APEX-Agents-AA
1	Gemini 3.5 Flash	high	47.1%
2	Kimi K3	max	41.3%
3	GPT-5.6 Terra	max	38.9%
4	GPT-5.5	xhigh	37.7%
5	GPT-5.6 Luna	max	35.8%
6	GLM 5.2	max	33.7%
7	GPT-5.4	xhigh	33.3%
8	Kimi K2.6		28.5%
9	GPT-5.4 Mini	xhigh	28.2%
10	Ling 3.0 Flash Fin		27.4%
11	GPT-5.4 Nano	xhigh	24.9%
12	DeepSeek V4 Pro 0424	max	24.3%
13	Qwen 3.7 Plus		22.4%
14	Grok 4.3	high	17.0%
15	Qwen 3.5 397B-A17B		15.3%
16	GLM 5		14.5%
17	Kimi K2.5		11.5%
18	MiniMax M2.7		10.6%
19	gpt-oss-120b	high	3.1%
20	MiMo V2.5 Pro		2.4%
21	gpt-oss-20b	high	0.7%