mLateOn measured on 9 of 261 Embeddings benchmarks9
BenchmarkByScoreRank
No benchmark matches
self-host onlyprice
57.6BEIR
Jul 2026released
8K tokensMax input
en, es, fr +6Languages
Alternatives?
ModelScoreCostPriceWhat changes
BetterQwen 3 Embedding 8BunstatedOpen+9%—$0.01Languages: not verified+1
OpenNV Embed v2unstated+10%——Languages: loses es, fr, de +5+1
BetterBge En IclunstatedOpen+8%—$0.01Languages: not verified+1BetterQwen 3 Embedding 4BunstatedOpen+7%—$0.02Languages: not verified+1VBetterVoyage 3 M Expunstated+18%——Languages: not verifiedModelScoreCostPriceWhat changes
Nothing here yet.
ModelScoreCostPriceWhat changes
Qwen 3 Embedding 8BunstatedOpen+9%—$0.01Languages: not verified+1Bge En IclunstatedOpen+8%—$0.01Languages: not verified+1Qwen 3 Embedding 4BunstatedOpen+7%—$0.02Languages: not verified+1VVoyage 3 M Expunstated+18%——Languages: not verifiedTConan Embedding v2unstated+15%——Languages: loses es, fr, de +5ModelScoreCostPriceWhat changes
NV Embed v2unstatedOpen+10%——Languages: loses es, fr, de +5+1Qwen 3 Embedding 8BunstatedOpen+9%—$0.01Languages: not verified+1Bge En IclunstatedOpen+8%—$0.01Languages: not verified+1yLENS D8000unstatedOpen+7%——Languages: not verified+1iInf Retriever v1unstatedOpen+7%——Languages: loses es, fr, de +5+1ModelScoreCostPriceWhat changes
lLateOnOpensame——Max input: 8K tokens → 0K tokens+1lmDenseOnOpensame——lDenseOnOpen-2%——Max input: 8K tokens → 1K tokens+1lColBERT ZeroOpen-4%——Max input: 8K tokens → 1K tokens+1lGTE ModernColBERT v1Open-5%——Max input: 8K tokens → 0K tokens+1lColBERT Zero SupervisedOpen-10%——Max input: 8K tokens → 1K tokens+1lColBERT Zero UnsupervisedOpen-11%——Max input: 8K tokens → 0K tokens+1lDenseOn UnsupervisedOpen-15%——Max input: 8K tokens → 1K tokens+1All lightonai 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
Benchmarks
in Embeddings & RetrievalBEIR · Retrieval · English57.6#24
FAQ
How much does mLateOn cost?›
There is no per-unit API price: self-host only.
Is mLateOn free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is mLateOn, and what is it best at?›
#24 of 194 in Embeddings.
Is mLateOn better than Stella En 400M v5?›
On BEIR · Mean over task types · English, mLateOn scores 57.6 vs 57.9 — about the same.
When was mLateOn released? Is there a newer version?›
Released 2026-07-30. It is the current version in its line.
Is mLateOn open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run mLateOn locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is mLateOn?›
mLateOn is an open-weights embeddings & retrieval model from lightonai, released Jul 2026.
What are the best alternatives to mLateOn?›
Qwen 3 Embedding 8B (better), NV Embed v2 (open weights).