Tau3 Banking: leaderboard
Artificial Analysis Tau3 banking-domain customer-service tasks, measuring agent success at policy-grounded banking support workflows.
Metric: Success Rate (%). Source: artificialanalysis.ai. Saturation forecast: Around September 2027. 209 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.8 Max | 51.34 | #35 |
| 2 | Grok 4.6 (High) | 50.72 | #26 (Grok 4.6) |
| 3 | Muse Spark 1.3 (Max) | 50.52 | #24 (Muse Spark 1.3) |
| 4 | GLM-5.3 (Max) | 50.31 | #39 (GLM-5.3) |
| 5 | Qwen3.8 2.4T A95B | 49.07 | #55 |
| 6 | Qwen 3.8 27B (xHigh) | 48.04 | #105 (Qwen 3.8 27B) |
| 7 | Qwen 3.8 Max (0902) | 47.84 | #29 |
| 8 | Qwen 3.8 27B (Medium) | 47.42 | #105 (Qwen 3.8 27B) |
| 9 | GLM-5.3 Flash | 47.22 | #65 |
| 10 | Muse Spark 1.3 (xHigh) | 47.22 | #24 (Muse Spark 1.3) |
| 11 | Kimi K3 (Max) | 45.98 | #30 (Kimi K3) |
| 12 | Gemini 3.8 Flash (Medium) | 45.77 | #21 (Gemini 3.8 Flash) |
| 13 | Qwen3.8-Flash-Next | 45.36 | #72 |
| 14 | Gemini 3.8 Flash (High) | 44.95 | #21 (Gemini 3.8 Flash) |
| 15 | Claude Opus 5 (High) | 44.74 | #12 (Claude Opus 5) |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=tau3-banking · How It Works · Data refreshed daily, snapshot 2026-10-20.