PARSA-Bench - Intent Detection: leaderboard
Metric: Macro-F1 (0-1 times 100) over the union of gold and predicted labels on 500 spoken Persian MASSIVE commands (56 intents, synthesized speech); zero-shot audio input, temperature 0, items with no extractable answer (including refusals) excluded; proprietary models on 100 items; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 8 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Flash | 79 | #237 |
| 2 | Gemma 3n E4B | 37 | #856 |
| 3 | Gemma 3n E2B | 24 | #1137 |
| 4 | Qwen2.5-Omni-7B | 12 | #601 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=parsa-bench-intent-detection · How It Works · Data refreshed daily, snapshot 2026-10-11.