Llama Nemotron Colembed VL 3B v2
Jan 2026
Llama Nemotron Colembed VL 3B v2Compare inon
Llama Nemotron Colembed VL 3B v2 measured on 5 of 203 Multimodal benchmarks5
BenchmarkByScoreRank
No benchmark matches
self-host onlyprice
88.7MIEB lite
Jan 2026released
8K tokensMax input
Ranking?
Full ranking →···
9Colqwen3.5 4.5B v383.710Llama Nemotron Colembed VL 3B v283.611Tomoro Colqwen3 Embed 8B83.5Strong at
Alternatives?
ModelScoreCostPriceWhat changes
DBetterArgus Colqwen3.5 9B v0unstatedOpen+3%——Memory (4-bit): 2.6 GB → 5.3 GBvOpenVultronRetrieverPrime Qwen3.5 8B+2%——Memory (4-bit): 2.6 GB → 5 GBOOpenOps Colqwen3 4B+1%——Memory (4-bit): 2.6 GB → 2.9 GBOpenNemotron Colembed VL 8B v2+1%——Memory (4-bit): 2.6 GB → 5.2 GBvOpenVultronRetrieverCore Qwen3.5 4.5B+1%——Memory (4-bit): 2.6 GB → 2.7 GBModelScoreCostPriceWhat changes
Nothing here yet.
ModelScoreCostPriceWhat changes
DArgus Colqwen3.5 9B v0unstatedOpen+3%——Memory (4-bit): 2.6 GB → 5.3 GBModelScoreCostPriceWhat changes
DArgus Colqwen3.5 9B v0unstatedOpen+3%——Memory (4-bit): 2.6 GB → 5.3 GBvVultronRetrieverPrime Qwen3.5 8BOpen+2%——Memory (4-bit): 2.6 GB → 5 GBOOps Colqwen3 4BOpen+1%——Memory (4-bit): 2.6 GB → 2.9 GBNemotron Colembed VL 8B v2Open+1%——Memory (4-bit): 2.6 GB → 5.2 GBvVultronRetrieverCore Qwen3.5 4.5BOpen+1%——Memory (4-bit): 2.6 GB → 2.7 GBModelScoreCostPriceWhat changes
Nemotron Colembed VL 8B v2Open+1%——Memory (4-bit): 2.6 GB → 5.2 GBNemotron Colembed VL 4B v2Opensame——Memory (4-bit): 2.6 GB → 2.9 GBLlama Nemoretriever Colembed 3B v1Opensame——Llama Nemoretriever Colembed 1B v1Open-1%——Nemotron 3.5 Embed 1B BF16Openn/a—$0.015Nemotron 3.5 Embed 8B BF16Openn/a—$0.035Memory (4-bit): 2.6 GB → 5 GBLlama Nemotron Embed VL 1B v2Openn/a—$0.01Omni Embed Nemotron 3BOpenn/a——Memory (4-bit): 2.6 GB → 2.8 GBAll NVIDIA models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · 8-bit✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
Licencehttps://huggingface.co/nvidia/llama-nemotron-colembed-vl-3b-v2/blob/main/LICENSEcommercial use with conditions
Benchmarks
in Embeddings & RetrievalFAQ
How much does Llama Nemotron Colembed VL 3B v2 cost?›
There is no per-unit API price: self-host only.
Is Llama Nemotron Colembed VL 3B v2 free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Llama Nemotron Colembed VL 3B v2, and what is it best at?›
#6 of 89 in Multimodal. Strongest at: Embeddings & Retrieval.
Is Llama Nemotron Colembed VL 3B v2 better than Colqwen3.5 4.5B v3?›
On ViDoRe v1&v2 · Mean over tasks · multilingual (4 languages), Llama Nemotron Colembed VL 3B v2 scores 83.6 vs 83.7 — about the same.
When was Llama Nemotron Colembed VL 3B v2 released? Is there a newer version?›
Released 2026-01-21. It is the current version in its line.
Is Llama Nemotron Colembed VL 3B v2 open source?›
Yes, the weights are open (https://huggingface.co/nvidia/llama-nemotron-colembed-vl-3b-v2/blob/main/LICENSE licence).
Can I run Llama Nemotron Colembed VL 3B v2 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is Llama Nemotron Colembed VL 3B v2?›
Llama Nemotron Colembed VL 3B v2 is an open-weights embeddings & retrieval model from NVIDIA, released Jan 2026.
What are the best alternatives to Llama Nemotron Colembed VL 3B v2?›
Argus Colqwen3.5 9B v0 (better).