τ²-Bench Telecom (AA)

Artificial Analysis' τ²-Bench telecom run. Shown for reference.

Published by

Artificial Analysis

Category

Agents & tools

Index weight

Reference only

Models

392

Data as of

9 Sept 2026

Source: Artificial Analysis (artificialanalysis.ai). Data taken from the public leaderboard.

1. 1GLM-5.299.1%
2. 2JT-35B-Flash99.1%
3. 3GLM-4.7-Flash98.8%
4. 4Claude Fable 598.5%
5. 5GLM-5-Turbo98.5%
6. 6GLM-5V-Turbo98.5%
7. 7Step 3.7 Flash98.5%
8. 8GLM-598.3%
9. 9GLM-5.197.7%
10. 10Qwen3.6 Plus97.7%
11. 11Grok 4.397.7%
12. 12GLM-597.4%
13. 13GLM-5.197.1%
14. 14Grok 4.2096.5%
15. 15DeepSeek V4 Pro96.2%

SearchAll providersAI21 LabsAlibabaAllen Institute for AIAllenaiAmazonAnt GroupAnt-groupAnthropicArcee AIBaiduByteDanceChina MobileCohereDeepSeekGoogleIBMInceptionKimiKorea TelecomKwaiKATLG AI ResearchLiquid AILongCatMBZUAI Institute of Foundation ModelsMetaMicrosoftMiniMaxMistral AIMoonshot AIMotif TechnologiesMultiverse ComputingNanbeigeNaverNex AGINous ResearchNVIDIAOpenAIOpenBMBPrime IntellectQwenReka AISarvamServiceNowStepFunSwiss AI InitiativeTII UAETrillion LabsUnknownUpstagexAIXiaomiZhipu AIOpen weights392 of 392

#	Model	Score ↓	Run ↕	Index ↕	Published ↕
1	GLM-5.2Zhipu AIopen	99.1%	GLM-5.2 (max)	57.4	–
2	CJT-35B-FlashChina Mobile	99.1%	–	52.1	–
3	GLM-4.7-FlashZhipu AIopen	98.8%	–	50.0	–
4	Claude Fable 5Anthropic	98.5%	Claude Fable 5 (with fallback)	70.4	–
5	GLM-5-TurboZhipu AI	98.5%	–	58.1	–
6	GLM-5V-TurboZhipu AI	98.5%	GLM 5V Turbo	57.2	–
7	Step 3.7 FlashStepFunopen	98.5%	Step 3.7 Flash	54.6	–
8	GLM-5Zhipu AIopen	98.3%	GLM-5	59.0	–
9	GLM-5.1Zhipu AIopen	97.7%	GLM-5.1	60.1	–
10	Qwen3.6 PlusAlibabaopen	97.7%	Qwen3.6 Plus	59.0	–
11	Grok 4.3xAI	97.7%	Grok 4.3 (high)	57.8	–
12	GLM-5Zhipu AIopen	97.4%	–	54.5	–
13	GLM-5.1Zhipu AIopen	97.1%	–	52.6	–
14	Grok 4.20xAI	96.5%	0309	57.9	–
15	DeepSeek V4 ProDeepSeekopen	96.2%	0424	59.7	–
16	Qwen3 6Alibaba	95.9%	–	62.0	–
17	Kimi K2.6Moonshot AIopen	95.9%	Kimi K2.6	60.6	–
18	Kimi K2.5Moonshot AIopen	95.9%	Kimi K2.5	56.1	–
19	GLM-4.7Zhipu AIopen	95.9%	GLM-4.7	54.9	–
20	Gemini 3.1 ProGoogle	95.6%	–	63.9	–
21	Gemini 3.5 FlashGoogle	95.6%	–	63.9	–
22	DeepSeek V4 FlashDeepSeekopen	95.6%	high, 0420	59.3	–
23	Qwen3.5 397B-A17BAlibabaopen	95.6%	Qwen3.5 397B A17B	58.6	–
24	Gemini 3.5 FlashGoogle	95.3%	Gemini 3.5 Flash	60.7	–
25	MiniMax-M2.5MiniMaxopen	95.3%	MiniMax-M2.5	53.5	–
26	Qwen3.6 35B-A3BAlibabaopen	95.3%	Qwen3.6 35B A3B	53.2	–
27	XMimo v2 ProXiaomi	95.0%	–	60.5	–
28	DeepSeek V4 FlashDeepSeekopen	95.0%	0420	59.1	–
29	XMiMo-V2-FlashXiaomiopen	95.0%	–	53.4	–
30	Qwen3 7Alibaba	94.7%	–	58.9	–
31	Claude Opus 4.8Anthropic	94.4%	Claude Opus 4.8 (max)	64.7	–
32	Step 3.5 FlashStepFunopen	94.4%	0202	54.9	–
33	DeepSeek V4 FlashDeepSeekopen	94.4%	–	48.4	–
34	DeepSeek V4 ProDeepSeekopen	94.2%	high, 0424	60.0	–
35	XMiMo-V2.5-ProXiaomiopen	94.2%	–	59.0	–
36	Qwen3.6 27BAlibabaopen	94.2%	Qwen3.6 27B	53.2	–
37	Mistral Medium 3.5Mistral AI	94.2%	Mistral Medium 3.5	50.8	–
38	GLM-4.7Zhipu AIopen	94.2%	–	50.4	–
39	GPT-5.5OpenAI	93.9%	GPT-5.5 (xhigh)	66.6	–
40	Qwen3.5 27BAlibabaopen	93.9%	Qwen3.5 27B	55.4	–
41	Kimi K2.6Moonshot AIopen	93.9%	–	55.3	–
42	Qwen3.5 122B-A10BAlibabaopen	93.6%	Qwen3.5 122B A10B	56.0	–
43	Qwen3.6 27BAlibabaopen	93.6%	–	50.7	–
44	Grok 4.1xAI	93.3%	–	55.9	–
45	XMiMo-V2-FlashXiaomiopen	93.3%	feb-2026	53.3	–
46	TTri-21B-ThinkTrillion Labsopen	93.3%	–	44.0	–
47	GPT-5.5OpenAI	93.0%	–	66.3	–
48	Qwen3.7 PlusAlibaba	93.0%	Qwen3.7 Plus	59.7	–
49	Kimi K2 ThinkingMoonshot AIopen	93.0%	–	52.8	–
50	CJT-MINIChina Mobile	93.0%	–	43.6	–
51	Nova 2.0 Pro PreviewAmazon	92.7%	–	52.3	–
52	ARing-2.6-1TAnt Group	92.4%	–	51.1	–
53	Claude Opus 4.6Anthropic	92.1%	–	63.1	–
54	GPT-5.2 CodexOpenAI	92.1%	GPT-5.2 Codex (xhigh)	60.7	–
55	Qwen3.5-4BAlibaba	92.1%	Qwen3.5 4B	47.1	–
56	GPT-5.5OpenAI	91.8%	–	64.9	–
57	GLM-4.7-FlashZhipu AIopen	91.8%	–	41.3	–
58	Muse SparkMeta	91.5%	–	65.4	–
59	Grok 4.3xAI	91.2%	–	60.7	–
60	XMimo v2 OmniXiaomi	91.2%	–	59.1	–

Show 100 more