A Matter of TASTE: leaderboard
As agent capabilities advance, existing benchmarks, such as τ^2-Bench, are becoming increasingly saturated. Yet constructing new benchmark tasks remains complex, costly, and labor-intensive. Moreover, the standard approach, in which scenarios are first written in natural language and then mapped to tool sequences, captures only a narrow subset of the tool-use patterns agents exercise. In this pa
Metric: Airline (Pass@1) τBV (self-reported). Source: benchmarklist.com. Saturation forecast: Around June 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 82 | #99 |
| 2 | Claude Sonnet 4.6 | 72 | #90 |
| 3 | Gemini 2.5 Flash | 66 | #245 |
| 4 | GPT-5.2 | 58 | #110 |
| 5 | DeepSeek V3.1 | 49 | #264 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=a-matter-of-taste · How It Works · Data refreshed daily, snapshot 2026-10-20.