All tasks / Comparisons

KaLM-Embedding-Gemma3-12B-2511 vs Qwen3-Embedding-8B

Compared only on public benchmark results, across the 1 task where both are listed: they share a tier in 1.

KaLM-Embedding-Gemma3-12B-2511

Tencent's open-weights 12B multilingual text embedding model built on Gemma 3.

Vendor: Tencent
Weights: Open: run it yourself
Released: Nov 2025
GPU memory to run locally: ~24 GB
License: Tencent KaLM Embedding Community License
Last verified: 19 Oct 2026

Alternatives to KaLM-Embedding-Gemma3-12B-2511

Qwen3-Embedding-8B

Alibaba's Apache-2.0 open-weights multilingual text embedding models in 0.6B, 4B and 8B sizes.

Vendor: Alibaba (Qwen)
Weights: Open: run it yourself
Released: Jun 2025
GPU memory to run locally: ~16 GB
License: Apache-2.0
Last verified: 19 Oct 2026

Alternatives to Qwen3-Embedding-8B

Task by task

Task	KaLM-Embedding-Gemma3-12B-2511	Qwen3-Embedding-8B	Higher tier
Embeddings & retrieval Models	S72.3% on MTEB(Multilingual, v2) · Local	S70.6% on MTEB(Multilingual, v2) · Local	Same tier

A tier spans a range of scores, and small gaps on a leaderboard are often within its margin of error, so only a different tier counts as ahead. How we rank.

Same benchmark, both results

Benchmark	KaLM-Embedding-Gemma3-12B-2511	Qwen3-Embedding-8B	Snapshot
MTEB(Multilingual, v2) Mean (Task): average main score over 131 multilingual tasks (bitext mining, classification, clustering, retrieval, reranking, STS, instruction reranking), higher is better	72.3% tencent/KaLM-Embedding-Gemma3-12B-2511	70.6% Qwen/Qwen3-Embedding-8B	19 Oct 2026