KURE v2 measured on 2 of 261 Embeddings benchmarks2
BenchmarkByScoreRank
No benchmark matches
self-host onlyprice
69.7MTEB Korean v2
Aug 2026released
8K tokensMax input
en, koLanguages
Alternatives?
ModelScoreCostPriceWhat changes
sOpenComsat Embed Ko 8B Preview-2%——Memory (4-bit): 0.1 GB → 4.5 GBModelScoreCostPriceWhat changes
Nothing here yet.
ModelScoreCostPriceWhat changes
Nothing here yet.
ModelScoreCostPriceWhat changes
sComsat Embed Ko 8B PreviewOpen-2%——Memory (4-bit): 0.1 GB → 4.5 GBModelScoreCostPriceWhat changes
nKURE v1Open-6%——Memory (4-bit): 0.1 GB → 0.3 GBnKoE5Open-10%——Max input: 8K tokens → 1K tokens+1All nlpai-lab models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceApache-2.0free for commercial use
History
2 versionsBenchmarks
in Embeddings & RetrievalFAQ
How much does KURE v2 cost?›
There is no per-unit API price: self-host only.
Is KURE v2 free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is KURE v2, and what is it best at?›
#2 of 150 in Embeddings.
Is KURE v2 better than Comsat Embed Ko 8B Preview?›
On MTEB Korean v2 · Retrieval · multilingual (2 languages), KURE v2 scores 81.1 vs 79.3 — higher.
When was KURE v2 released? Is there a newer version?›
Released 2026-08-17. It is the current version in its line.
Is KURE v2 open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run KURE v2 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is KURE v2?›
KURE v2 is an open-weights embeddings & retrieval model from nlpai-lab, released Aug 2026.
What are the best alternatives to KURE v2?›
Comsat Embed Ko 8B Preview (open weights).