Benchmarks · agentic
τ²-Bench Telecom
τ²-Bench telecom domain: dual-control tool-using agent tasks; pass rate.
96 models · 156 results · data as of 2026-10-29 · artificialanalysis.ai ↗
Leaderboard
GLM 5.2 max
More from this benchmark
3 published measuresScore vs any published measure3 measures: sub-scores, splits, counts
Breakdown
darker is better within each column · click a column to sort| Model | |||
|---|---|---|---|
| GLM 5.2 max | 23,951,355 | 257,562 | 167,797 |
| GLM 4.7 Flash | 23,066,047 | 237,272 | 157,482 |
| GLM 5 | 25,245,081 | 249,956 | 178,584 |
| GLM 5.1 | 24,789,372 | 258,997 | 173,224 |
| Grok 4.3 high | 25,120,183 | 176,339 | 247,917 |
| Qwen 3.6 Plus | 26,547,437 | 264,129 | 317,388 |
| DeepSeek V4 Pro 0424 max | 27,043,460 | 242,785 | 251,324 |
| GLM 4.7 | 24,205,743 | 241,809 | 11,452 |
| Kimi K2.5 | 22,537,488 | 251,372 | 173,991 |
| Kimi K2.6 | 22,385,132 | 222,466 | 317,358 |
| DeepSeek V4 Flash 0420 high | 28,501,687 | 247,496 | 180,259 |
| Gemini 3.5 Flash medium | 30,130,653 | 201,627 | 525,235 |
| Qwen 3.5 397B-A17B | 27,671,109 | 295,455 | 210,486 |
| MiniMax M2.5 | 24,603,564 | 249,928 | 191,208 |
| Qwen 3.6 35B-A3B | 25,860,679 | 253,867 | 447,676 |
| MiMo V2 Pro | 26,994,466 | 273,792 | 281,400 |
| Qwen 3.7 Max | 26,458,210 | 250,334 | 133,537 |
| Opus 4.8 max | 39,964,734 | 330,522 | 940,972 |
| MiMo V2.5 Pro | 27,169,095 | 273,958 | 286,846 |
| Mistral Medium 3.5 high | 29,065,398 | 173,552 | 312,543 |
| Qwen 3.6 27B | 25,085,232 | 237,035 | 369,654 |
| GPT-5.5 xhigh | 20,437,448 | 164,648 | 302,225 |
| Qwen 3.5 27B | 27,349,047 | 270,216 | 156,769 |
| Qwen 3.5 122B-A10B | 28,094,459 | 270,605 | 182,757 |
| Qwen 3.7 Plus | 27,941,723 | 265,844 | 141,617 |
All results
156 rows| # | Model | Effort | τ²-Bench Telecom |
|---|---|---|---|
| 1 | GLM 5.2 | max | 99.1% |
| 2 | GLM 4.7 Flash | 98.8% | |
| 3 | GLM 5 | 98.2% | |
| 4 | GLM 5.1 | 97.7% | |
| 5 | Grok 4.3 | high | 97.7% |
| 6 | Qwen 3.6 Plus | 97.7% | |
| 7 | GLM 5 | none | 97.4% |
| 8 | GLM 5.1 | none | 97.1% |
| 9 | DeepSeek V4 Pro 0424 | max | 96.2% |
| 10 | GLM 4.7 | 95.9% | |
| 11 | Kimi K2.5 | 95.9% | |
| 12 | Kimi K2.6 | 95.9% | |
| 13 | DeepSeek V4 Flash 0420 | high | 95.6% |
| 14 | Gemini 3.5 Flash | medium | 95.6% |
| 15 | Qwen 3.5 397B-A17B | 95.6% | |
| 16 | Gemini 3.5 Flash | high | 95.3% |
| 17 | MiniMax M2.5 | 95.3% | |
| 18 | Qwen 3.6 35B-A3B | 95.3% | |
| 19 | DeepSeek V4 Flash 0420 | max | 95.0% |
| 20 | MiMo V2 Pro | 95.0% | |
| 21 | Qwen 3.7 Max | max | 94.7% |
| 22 | Opus 4.8 | max | 94.4% |
| 23 | DeepSeek V4 Flash 0420 | none | 94.4% |
| 24 | DeepSeek V4 Pro 0424 | high | 94.2% |
| 25 | GLM 4.7 | none | 94.2% |
| 26 | MiMo V2.5 Pro | 94.2% | |
| 27 | Mistral Medium 3.5 | high | 94.2% |
| 28 | Qwen 3.6 27B | 94.2% | |
| 29 | GPT-5.5 | xhigh | 93.9% |
| 30 | Kimi K2.6 | none | 93.9% |