Claim Detection on Multilingual Fact-Checking (114 languages) (test)
1.3Median Latency (ms)XLM-R-Large (FT)
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| XLM-R-Large (FT)Hosting=self-hosted, Evaluation Protocol=Fine-tuned2026.06 | 1.3 | 1.3 | 722.5 | 1 | 72.77 | 90.45 | 0.002 | |
| Qwen3-8bHosting=self-hosted2026.06 | 135.1 | 145.2 | 7.4 | 101 | 47.42 | 36.53 | 0.0289 | |
| GPT-5.2 (API)Hosting=API, Evaluation Protocol=API2026.06 | 413 | 790 | 2.4 | 318 | 67.92 | 61.79 | 0.25 | |
| Claude Opus 4.6 (API)Hosting=API, Evaluation Protocol=API2026.06 | 1,735 | 5,335 | 0.6 | 1,335 | 74.89 | 68.86 | 4.35 |