𝜏²-Bench Telecom

에이전트와 사용자가 통신 서비스 문제를 해결하기 위해 행동을 조율해야 하는 기술 지원 시나리오를 모의하는 이중 제어 대화형 AI 벤치마크입니다.

모델: 147 현재, 440 전체
중앙값: 36.3%
범위: 0.0%–99.1%

점수 분포

가장 낮음: 0.0%
중앙값: 36.3%
최고: 99.1%

리더보드

지원 중단된 모델 포함

순위	모델	점수
1	JT-35B-Flash China Mobile	99.1%
2	Gemini 3.1 Pro Preview Google	95.6%
3	Qwen3.5 397B A17B (Reasoning) Alibaba	95.6%
4	Qwen3.6 35B A3B (Reasoning) Alibaba	95.3%
5	Mistral Medium 3.5 Mistral	94.2%
6	Qwen3.5 122B A10B (Reasoning) Alibaba	93.6%
7	JT-MINI China Mobile	93.0%
8	Qwen3.7 Plus Alibaba	93.0%
9	Nova 2.0 Pro Preview (Medium) Amazon	92.7%
10	Ring-2.6-1T InclusionAI	92.4%
11	Qwen3.5 4B (Reasoning) Alibaba	92.1%
12	Nova 2.0 Pro Preview (Low) Amazon	90.6%
13	Kimi K2.7 Code Kimi	90.1%
14	Trinity Large Thinking Arcee AI	90.1%
15	KAT Coder Pro V2 KwaiKAT	89.5%
16	MiniMax-M3 MiniMax	88.9%
17	Qwen3.5 Omni Plus Alibaba	88.3%
18	MiniCPM-V 4.6 1.3B OpenBMB	87.7%
19	Qwen3.5 4B (Non-reasoning) Alibaba	87.7%
20	HyperCLOVA X SEED Think (32B) Naver	87.4%
21	Qwen3.5 9B (Reasoning) Alibaba	86.8%
22	Mi:dm K 2.5 Pro Korea Telecom	86.5%
23	GPT-5.3 Codex (Xhigh) OpenAI	86.0%
24	Qwen3.5 9B (Non-reasoning) Alibaba	85.1%
25	Qwen3.6 35B A3B (Non-reasoning) Alibaba	85.1%
26	Qwen3.5 122B A10B (Non-reasoning) Alibaba	84.5%
27	Qwen3.5 Omni Flash Alibaba	84.5%
28	Qwen3.5 397B A17B (Non-reasoning) Alibaba	83.9%
29	ERNIE 5.0 Thinking Preview Baidu	83.9%
30	Nemotron 3 Ultra 550B A55B (Reasoning) NVIDIA	83.3%
31	MiniCPM5-1B (Non-reasoning) OpenBMB	82.5%
32	Nex-N2-Pro (Based on Qwen3.5-397B-A17B) Nex AGI	81.6%
33	Qwen3.5 2B (Non-reasoning) Alibaba	81.6%
34	MiniCPM5-1B (Reasoning) OpenBMB	81.0%
35	Tri-21B-Think Trillion Labs	81.0%
36	Command A+ Cohere	80.7%
37	o3 OpenAI	80.7%
38	Nova 2.0 Omni (Medium) Amazon	80.4%
39	LongCat Flash Lite LongCat	79.5%
40	Qwen3 Coder Next Alibaba	79.5%
41	EXAONE 4.5 33B (Reasoning) LG AI Research	78.1%
42	Nova 2.0 Lite (Medium) Amazon	75.7%
43	Nova 2.0 Lite (High) Amazon	72.8%
44	Nova 2.0 Lite (Low) Amazon	71.9%
45	Nova 2.0 Pro Preview (Non-reasoning) Amazon	71.6%
46	Apriel-v1.6-15B-Thinker ServiceNow	69.3%
47	Qwen3.5 2B (Reasoning) Alibaba	69.0%
48	Nova 2.0 Omni (Low) Amazon	67.8%
49	gpt-oss-120b (High) OpenAI	65.8%
50	Gemma 4 31B (Non-reasoning) Google	65.5%