MTEB(Multilingual) mean (task)
We haven't written a description for this benchmark yet; see the card notes below for how each maker ran it.
- ModelScoreYour cost / mo
- 1Qwen3-Embedding-8B · Alibaba70.58 score
- 2Gemini Embedding 2 · Google69.9 score
“Best value”: no other model here reports a higher score at a lower cost for your workload. Scores are self-reported by each maker with different settings, so treat close calls as ties.
How each maker reported it
- Qwen3-Embedding-8B technical report Tables 2–3; model card
Alibaba's own numbers. Competitor scores were taken from the MTEB leaderboard in early June 2025. CMTEB for the 8B is 73.84 in the model card and Table 8 vs 73.83 in Table 3.
- Gemini Embedding 2 technical report, Tables 1–2
Google's own numbers. Competitor scores are partly measured by Google through public APIs, partly self-reported. Table 1 overall mean is over tasks where all four models have scores. 'Gemini Embedding (previous)' is the earlier text-only model. Nova MME reported only the MTEB(Multilingual) task mean. Native-audio MSEB passage retrieval: 73.99 MRR@10 vs 70.40 with ASR transcripts (Table 3).
Only each model's own maker's number is shown; scores a maker reports for rival models are left out.