Classifica

Classifica: Chiamata di strumenti

Chiamate di funzioni e orchestrazione API nei flussi agentici: Toolathlon, MCP Atlas e τ²-Bench.

Generale767 Coding469 Matematica90 Reasoning483 Agentico89 Conoscenza482 Multimodale452 Agenti di codice68 Reasoning scientifico96 Multilingue12 Scrittura95 Chat e assistenti95 Ricerca99 Chiamata di strumenti45 Contesto lungo116 Velocità472 Valore471 Open weights183

45 modelli

#	Modello	Podium Score	Toolathlon	MCP Atlas	τ²-Bench Retail	Velocità
01	Claude Fable 5 Anthropicreasoning	98.5	—	—	98.5%	71 t/s
02	Gemini 3.5 Flash Google	95.3	—	—	95.3%	267 t/s
03	MiMo V2.5 Pro XiaomiAperto	94.2	—	—	94.2%	65 t/s
4	Claude 5.5 Sonnet Anthropicreasoning	92.5	—	—	92.5%	94 t/s
5	Muse Spark Meta	91.5	—	—	91.5%	80 t/s
6	Kimi K2.7 Code Moonshot AIAperto	90.1	—	—	90.1%	39 t/s
7	Kimi K3.1 Pro Moonshot AIApertoreasoning	90.0	—	—	90%	45 t/s
8	Gemini 4 Argon Googlereasoninganteprima	90.0	—	—	90%	88 t/s
9	GPT-6.1 Sol OpenAIreasoning	89.0	—	—	89%	75 t/s
10	MiniMax M3 MiniMax	88.9	—	—	88.9%	93 t/s
11	GPT-6 Astra OpenAIreasoning	88.0	—	—	88%	82 t/s
12	Gemini 3 Pro Google	87.1	—	—	87.1%	120 t/s
13	Gemini 3.1 Pro Preview Googlereasoninganteprima	86.0	—	—	86%	105 t/s
14	GPT-5.3 Codex OpenAI	86.0	—	—	86%	118 t/s
15	Qwen3.7 Max Alibaba	85.6	—	76.4%	94.7%	203 t/s
16	Kimi K3.1 Highspeed Moonshot AIApertoreasoning	84.0	—	—	84%	130 t/s
17	Mistral Large 4 Mistralreasoning	84.0	—	—	84%	92 t/s
18	Nemotron 3 Ultra 550B NVIDIAAperto	83.3	—	—	83.3%	142 t/s
19	MiMo v2.6 Flash XiaomiAperto	83.0	—	—	83%	118 t/s
20	Gemini 3.1 Pro Google	82.4	—	69.2%	95.6%	136 t/s
21	Claude 5.5 Haiku Anthropicreasoning	81.0	—	—	81%	145 t/s
22	NVIDIA Nemotron 3.5 Super-VL NVIDIAAperto	80.0	—	—	80%	110 t/s
23	Claude Opus 4.8 Anthropic	78.8	59.9%	82.2%	94.4%	50 t/s
24	Kimi K3 Moonshot AIApertoreasoning	78.7	73.2%	84.2%	—	37 t/s
25	Simba 3.2 Independentreasoning	78.0	—	—	78%	62 t/s
26	Claude Opus 4.7 Anthropic	77.3	—	77.3%	—	80 t/s
27	GLM-4.5 Air Zhipu AIApertoreasoning	76.0	—	—	76%	96 t/s
28	Gemini 3.1 Flash Lite Preview Googleanteprima	75.0	—	—	75%	210 t/s
29	GPT-5.5 OpenAI	74.9	55.6%	75.3%	93.9%	60 t/s
30	GLM-5.2 Z.aiAperto	74.7	48.2%	76.8%	99.1%	164 t/s
31	Claude Opus 4.6 Anthropic	73.8	—	62.7%	84.8%	50 t/s
32	Qwen3.7 Plus Alibaba	73.2	—	73.2%	—	54 t/s
33	DeepSeek V4 Pro DeepSeekAperto	72.2	51.8%	73.6%	91.2%	66 t/s
34	GPT-5.6 Sol OpenAI	71.6	58%	—	85.1%	72 t/s
35	Qwen3.6 Plus Alibaba	70.5	39.8%	74.1%	97.7%	55 t/s
36	GPT-5.6 Terra OpenAI	69.7	53.1%	—	86.3%	127 t/s
37	Hunyuan Hy3 Tencent	63.8	48.5%	79.1%	—	69 t/s
38	Claude Sonnet 4.6 Anthropic	61.3	—	61.3%	—	50 t/s
39	GPT-5.4 OpenAI	60.9	54.6%	67.2%	—	60 t/s
40	DeepSeek V4 Flash DeepSeekAperto	58.4	47.8%	69%	—	113 t/s
41	GPT-5.2 OpenAI	53.5	46.3%	60.6%	—	90 t/s
42	GPT-5.6 Luna OpenAI	53.4	53.4%	—	—	175 t/s
43	Gemini 3 Flash Google	53.4	49.4%	57.4%	—	120 t/s
44	Kimi K2.6 Moonshot AIAperto	50.0	50%	—	—	40 t/s
45	Qwen3.5 397B-A17B AlibabaAperto	38.3	38.3%	—	—	71 t/s

Il Podium Score è una combinazione ponderata di cinque classifiche. Consulta la Metodologia per i dettagli.

Model Name

Provider

Podium Score0.0

Context128K

Price (1M out)$0/M

Speed0 t/s

Architecture & Highlights

Model overview and workload profile.

Quickstart cURL

curl https://api.openrouter.ai/api/v1/chat/completions ...