Hear2Act (Text, Explicit Feedback): leaderboard
Hear2Act (2026) runs task-oriented assistants through 480 persona-grounded consumer-service scenarios with a hidden user concern and verifiable outcomes. On this board the concern is stated in the user words and a text LLM reads the transcript; the score is the percent of dialogues ending with the best option, higher is better.
Metric: Optimal-solution rate (%; share of dialogues ending with the first-tier option, using the final recommendation if the turn budget runs out; 480 persona-grounded consumer-service scenarios with a hidden user concern and verifiable outcomes; explicit lexical feedback: the concern is stated in the user's words; text LLM reading the dialogue transcript only, two runs per scenario). Source: arxiv.org. Saturation forecast: Around February 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 74.9 | #102 |
| 2 | Kimi K2.5 | 69.4 | #256 |
| 3 | GLM-5 | 67.3 | #260 |
| 4 | Qwen 3 32B | 64.3 | #960 |
| 5 | DeepSeek V3.2 | 54.8 | #406 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=hear2act-text-explicit-feedback · How It Works · Data refreshed daily, snapshot 2026-11-13.