AI RACE— A Corrida da IA
← Todos os rankings

🤖 AI Agents (Autonomous AI)

An AI's ability to autonomously execute multi-step actions to complete a task — beyond just answering questions. For example, given "book flights + hotel for a weekend trip to Da Nang, budget under 5 million" — the AI must search flights, compare prices, fill forms, and resolve errors on its own without step-by-step guidance. Measured via τ²-bench (dialogue + tool use) and Terminal-Bench v2 (multi-step terminal/computer tasks).

100/100 modelos
#Acesso
199.3Testar agora
298.0Testar agora
3
China Mobile
97.1–
496.9Testar agora
5
Anthropic
96.5Testar agora
6
SpaceXAI
96.1–
795.9Testar agora
895.8Testar agora
995.4Testar agora
1094.6Testar agora
11
Z AI
94.5Testar agora
1294.3Testar agora
1394.3Testar agora
1493.9Testar agora
1593.8Testar agora
1693.7Testar agora
1793.6Testar agora
1892.6Testar agora
1992.5–
2092.5Testar agora
2192.3Testar agora
2292.1Testar agora
2392.0Testar agora
2491.5Testar agora
2591.4Testar agora
26
China Mobile
91.3–
27
91.2–
2891.2Testar agora
29
OpenAI
91.1Testar agora
3091.1–
3191.1Testar agora
3291.0Testar agora
33
90.2–
3490.2Testar agora
3589.8Testar agora
3689.2–
37
SpaceXAI
89.2Testar agora
3889.1Testar agora
39
InclusionAI
88.7–
40
88.6–
4188.5Testar agora
4288.3Testar agora
4388.2Testar agora
4488.2Testar agora
4588.1Testar agora
4688.0Testar agora
4788.0Testar agora
4887.5Testar agora
4987.1Testar agora
5086.9Testar agora
5186.8Testar agora
5286.6–
5386.3Testar agora
5485.8Testar agora
5585.7Testar agora
5685.6Testar agora
5785.4Testar agora
58
OpenAI
84.5Testar agora
5984.4Testar agora
6084.3–
6183.9Testar agora
6283.7Testar agora
63
Motif Technologies
83.6–
6483.4Testar agora
6583.3Testar agora
66
Korea Telecom
83.3–
6783.1Testar agora
6883.1Testar agora
69
83.1–
7082.7–
7182.7Testar agora
72
MiniMax
82.6Testar agora
7382.5Testar agora
7482.4–
7582.1–
7682.0Testar agora
7781.9Testar agora
78
OpenAI
81.8Testar agora
7981.7Testar agora
8081.7Testar agora
8181.5Testar agora
82
Institute of Foundation Models
81.2–
83
MiniMax
80.9Testar agora
8480.7Testar agora
8580.4Testar agora
8680.3Testar agora
87
Xiaomi
80.2–
88
Trillion Labs
79.7–
8979.7Testar agora
9079.6Testar agora
91
79.1–
92
Nex AGI
79.0–
9378.6Testar agora
94
Multiverse Computing
78.3–
9578.2–
9678.1Testar agora
9777.7Testar agora
9877.6Testar agora
9975.7Testar agora
10075.2Testar agora

Como funciona a pontuação — AIM Index

● Capacidade 85%● Recência 15%
τ²-benchTerminal-Bench v2

Fontes de dados (1)