🤖 Classifica: Agentico

Uso autonomo di strumenti: OSWorld, Toolathlon, MCP Atlas, τ²-Bench.

🏆 Generale673💻 Coding392📐 Matematica32🧠 Reasoning391🤖 Agentico31📚 Conoscenza388👁️ Multimodale370🛠️ Agenti di codice40🔬 Reasoning scientifico45🌍 Multilingue12✍️ Scrittura53💬 Chat e assistenti53🔭 Ricerca48🔧 Chiamata di strumenti31📜 Contesto lungo55⚡ Velocità378💰 Valore377🔓 Open weights153
Fornitore
Licenza
31 modelli
#ModelloPodium ScoreOSWorldToolathlonMCP Atlasτ²-Bench RetailApex AgentsVelocità
01
Kimi K3
Moonshot AI · Aperto
100.0
—73.2%84.2%—37.6%37 t/s
02
96.2
———98.5%—71 t/s
03
75.9
———95.3%—267 t/s
4
74.9
—59.9%82.2%94.4%—50 t/s
5
74.3
——77.3%——80 t/s
6
71.5
——76.4%94.7%—203 t/s
7
MiMo V2.5 Pro
Xiaomi · Aperto
69.0
———94.2%—65 t/s
8
GLM-5.2
Z.ai · Aperto
66.9
—48.2%76.8%99.1%—164 t/s
9
62.6
——69.2%95.6%33.5%136 t/s
10
GPT-5.5
OpenAI
61.1
—55.6%75.3%93.9%—60 t/s
11
59.0
——73.2%——54 t/s
12
52.6
—39.8%74.1%97.7%—55 t/s
13
51.9
———91.5%—80 t/s
14
DeepSeek V4 Pro
DeepSeek · Aperto
49.7
—51.8%73.6%91.2%—66 t/s
15
43.3
—53.4%———175 t/s
16
43.1
72.7%—62.7%84.8%—50 t/s
17
Kimi K2.7 Code
Moonshot AI · Aperto
43.0
———90.1%—39 t/s
18
GPT-5.4
OpenAI
41.6
—54.6%67.2%——60 t/s
19
36.7
—48.5%79.1%—25.6%69 t/s
20
MiniMax M3
MiniMax
35.4
———88.9%—93 t/s
21
DeepSeek V4 Flash
DeepSeek · Aperto
35.3
—47.8%69%——113 t/s
22
33.9
—58%—85.1%—72 t/s
23
30.7
—53.1%—86.3%—127 t/s
24
Kimi K2.6
Moonshot AI · Aperto
26.3
—50%——27.9%40 t/s
25
24.1
———87.1%—120 t/s
26
GPT-5.2
OpenAI
17.4
—46.3%60.6%——90 t/s
27
17.1
———86%—118 t/s
28
15.9
—49.4%57.4%——120 t/s
29
7.3
72.5%—61.3%——50 t/s
30
Qwen3.5 397B-A17B
Alibaba · Aperto
0.0
—38.3%———71 t/s
31
Nemotron 3 Ultra 550B
NVIDIA · Aperto
0.0
———83.3%—142 t/s
Il Podium Score è una combinazione ponderata di cinque classifiche. Consulta la Metodologia per i dettagli.