Category ranking · 10 models · checked 2026-10-13
Embeddings and rerankers
Ranked by MMTEB (multilingual) mean task score; rerankers: MTEB-R. Source: Qwen3-Embedding and Qwen3-Reranker model cards (compared models taken from the MTEB leaderboard, 2025-05-24).
Embedding models (MMTEB mean task)
4 with a published score, 3 not yet measured.
# | Model | Parameters | Score | Source | Licence |
|---|---|---|---|---|---|
| 1 | Qwen3 Embedding 8B Alibaba Qwen | 7.6 B | 70.58Sourced | Source MTEB (Multilingual), mean (task). | Apache-2.0 |
| 2 | Qwen3 Embedding 0.6B Alibaba Qwen | 596 M | 64.33Sourced | Source MTEB (Multilingual), mean (task), Qwen3-Embedding card table. | Apache-2.0 |
| 3 | multilingual-e5-large-instruct Microsoft (intfloat) | 560 M | 63.22Sourced | Source Row in the Qwen3-Embedding card table (taken from the MTEB leaderboard, 2025-05-24). | MIT |
| 4 | BGE-M3 BAAI | — | 59.56Sourced | Source Row in the Qwen3-Embedding card table (taken from the MTEB leaderboard, 2025-05-24). | MIT |
| — | EmbeddingGemma 300M Google | 303 M | Not yet measured | Model card is gated; not read. | Gemma Terms of Use |
| — | jina-embeddings-v3 Jina AI | 572 M | Not yet measured | — | CC-BY-NC-4.0 Check terms before commercial use |
| — | Nomic Embed Text v1.5 Nomic AI | 137 M | Not yet measured | — | Apache-2.0 |
Rerankers (MTEB-R)
2 with a published score, 1 not yet measured.
# | Model | Parameters | Score | Source | Licence |
|---|---|---|---|---|---|
| 1 | Qwen3 Reranker 0.6B Alibaba Qwen | 596 M | 65.8Sourced | Source MTEB-R, model card table (reranker metric; not comparable with embedding scores). | Apache-2.0 |
| 2 | BGE Reranker v2 M3 BAAI | 568 M | 57.03Sourced | Source MTEB-R, row in the Qwen3-Reranker card table (reranker metric). | Apache-2.0 |
| — | mxbai-rerank-large v2 Mixedbread | 1.5 B | Not yet measured | — | Apache-2.0 |
How to read this: one public benchmark per category, so you can compare like with like. We copy the published number and link it; we do not re-run it, average different benchmarks or fill gaps. Vendor-reported numbers use each vendor's own settings, so treat small gaps as ties. “Not yet measured” means we found no comparable public score, not that the model is weak.
Every model was checked against the Hugging Face API on 2026-10-13: it exists, is public, and has the parameter count and licence shown. Other categories: Math and reasoning · Coding · Vision-language · OCR and documents · Speech to text · Text to speech · Image generation · Small fast 'System 1' · Biomedical · Tool use and agents · Multilingual and translation · Time series · Robotics (VLA).
Machine-readable: /api/v1/categories/embeddings-rerankers.json