SkMTEB - STS: leaderboard
SkMTEB is the Slovak Massive Text Embedding Benchmark: 31 datasets across seven task types (bitext mining, classification, clustering, pair classification, reranking, retrieval, semantic textual similarity). This board reports Mean score (%) over the two semantic-textual-similarity datasets; higher is better.
Metric: Mean score (%) over the two semantic-textual-similarity datasets; higher is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 28 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | jina-embeddings-v3 | 89.82 | |
| 2 | multilingual-e5-large-instruct | 88.86 | |
| 3 | gemini-embedding-001 | 88.74 | |
| 4 | paraphrase-multilingual-mpnet-base-v2 | 87.44 | |
| 5 | Qwen3-Embedding-8B | 86.54 | |
| 6 | bge-m3 | 86.39 | |
| 7 | multilingual-e5-large | 85.9 | |
| 8 | Qwen3-Embedding-4B | 85.82 | |
| 9 | slovakbert-sts-stsb | 85.69 | |
| 10 | jina-embeddings-v4 | 85.65 |
Interactive version: theaggregate.ai/benchmark?slug=skmteb-sts · How It Works · Data refreshed daily, snapshot 2026-11-13.