IntentGrasp - Toxic Speech: leaderboard
The Toxic Speech domain of the IntentGrasp (2026) All Set, intent-understanding items drawn from public intent datasets and posed as multiple choice with one or more correct intents. The score is instance F1 in percent, averaged over option-order shuffles; higher is better.
Metric: Instance F1 (%) on the Toxic Speech (TS) domain of the 12,909-item All Set, mean over option-order shuffles, zero-shot at temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 11 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B (IT) [4bit] | 55.41 | #1803 |
| 2 | Qwen3-32B [4bit] | 50.5 | #1273 |
| 3 | gemma-4-E4B-it | 49.94 | #1539 |
| 4 | Olmo-3.1-32B-Instruct [4bit] | 48.1 | |
| 5 | Gemma 4 E2B (IT) | 47.35 | #2451 |
| 6 | Llama 3.3 70B Instruct [4bit] | 44.11 | #1257 |
| 7 | Qwen 3 4B | 42.31 | #2348 |
| 8 | Qwen 3 8B | 38.97 | #1873 |
| 9 | OLMo 3 7B Instruct | 37.24 | #6953 |
| 10 | Llama 3.1 8B Instruct | 29.29 | #4600 |
| 11 | Llama 3.2 3B Instruct | 24.48 | #7241 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=intentgrasp-toxic-speech · How It Works · Data refreshed daily, snapshot 2026-11-21.