Colqwen2.5 v0.2 measured on 8 of 195 Multimodal benchmarks8
BenchmarkByScoreRank
No benchmark matches
self-host onlyprice
86.1MIEB lite
Jan 2025released
128K tokensMax input
enLanguages
Ranking?
Full ranking →···
30Qwen 3 VL Embedding 2B52.631Colqwen2.5 v0.251.932ColUltraVBERT49.3Strong at
Alternatives?
ModelScoreCostPriceWhat changes
BetterNemotron 3.5 Embed 1B BF16Open+11%—$0.015Max input: 128K tokens → 33K tokens
OpenEVIE 8B+28%——Languages: not verified+1BetterNemotron 3.5 Embed 8B BF16Open+16%—$0.035Max input: 128K tokens → 33K tokens+1
BetterQwen 3 VL Embedding 8BOpen+11%—$0.09Max input: 128K tokens → 33K tokens+2BetterEVIE 4.5BOpen+27%——Languages: not verified+1ModelScoreCostPriceWhat changes
Nothing here yet.
ModelScoreCostPriceWhat changes
Nemotron 3.5 Embed 1B BF16Open+11%—$0.015Max input: 128K tokens → 33K tokensNemotron 3.5 Embed 8B BF16Open+16%—$0.035Max input: 128K tokens → 33K tokens+1Qwen 3 VL Embedding 8BOpen+11%—$0.09Max input: 128K tokens → 33K tokens+2EVIE 8BOpen+28%——Languages: not verified+1EVIE 4.5BOpen+27%——Languages: not verified+1ModelScoreCostPriceWhat changes
EVIE 8BOpen+28%——Languages: not verified+1EVIE 4.5BOpen+27%——Languages: not verified+1wWebAI ColVec1.1 8BOpen+25%——Languages: not verified+1
pplx Embed v2 Late 9BOpen+25%——Max input: 128K tokens → 4K tokens+2vVultronRetrieverPrime Qwen3.5 8BOpen+24%——Memory (4-bit): 1.8 GB → 5 GBModelScoreCostPriceWhat changes
vColqwen2 v1.0Open-14%——Max input: 128K tokens → 33K tokensvColpali v1.3Open-17%——Max input: 128K tokens → 16K tokensvColSmol 256MOpen-59%——Max input: 128K tokens → 8K tokensvColSmol 500MOpenn/a——Max input: 128K tokens → 8K tokensvColpali v1.2Openn/a——Max input: 128K tokens → 16K tokensvColpali v1.1Openn/a——Max input: 128K tokens → 16K tokensAll vidore 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 Colqwen2.5 v0.2 cost?›
There is no per-unit API price: self-host only.
Is Colqwen2.5 v0.2 free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Colqwen2.5 v0.2, and what is it best at?›
#27 of 89 in Multimodal.
Is Colqwen2.5 v0.2 better than Qwen 3 VL Embedding 2B?›
On ViDoRe v3 · Mean over tasks · multilingual (6 languages), Colqwen2.5 v0.2 scores 51.9 vs 52.6 — about the same.
When was Colqwen2.5 v0.2 released? Is there a newer version?›
Released 2025-01-31. It is the current version in its line.
Is Colqwen2.5 v0.2 open source?›
Yes, the weights are open (Apache-2.0 licence).
Can I run Colqwen2.5 v0.2 locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
What is Colqwen2.5 v0.2?›
Colqwen2.5 v0.2 is an open-weights embeddings & retrieval model from vidore, released Jan 2025.
What are the best alternatives to Colqwen2.5 v0.2?›
Nemotron 3.5 Embed 1B BF16 (better), EVIE 8B (open weights).