SAVRN Model Hub · Comparisons
bge-small-zh-v1.5 vs jina-embeddings-v2-small-en
Bge-small-zh-v1.5 has 24M parameters and jina-embeddings-v2-small-en has 33M parameters; bge-small-zh-v1.5 is released under MIT License and jina-embeddings-v2-small-en under Apache License 2.0; at 16-bit, bge-small-zh-v1.5 needs about 0.1 GB (1x MI300X from $1.85 an hour) and jina-embeddings-v2-small-en about 0.1 GB (1x MI300X from $1.85 an hour).
| Field | bge-small-zh-v1.5 BAAI/bge-small-zh-v1.5 | jina-embeddings-v2-small-en jinaai/jina-embeddings-v2-small-en |
|---|---|---|
| Publisher | Beijing Academy of Artificial Intelligence | Jina AI |
| Task | Feature extraction | Feature extraction |
| Modality | Text | Text |
| Parameters, as reported | 24M parameters | 33M parameters |
| Architecture | BertModel | JinaBertForMaskedLM |
| Library | transformers | sentence-transformers |
| Context length | 512 tokens | 8,192 tokens |
| Repository size | 192.3 MB | 523.0 MB |
| Artifact formats | safetensors, pytorch | safetensors, onnx, pytorch, coreml |
| License | mit | apache-2.0 |
| Access | Open weights, no gate | Open weights, no gate |
| Memory at 16-bit (weights and margin) | 0.1 GB | 0.1 GB |
| Cheapest GPUs at 16-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Memory at 4-bit (weights and margin) | 0 GB | 0 GB |
| Cheapest GPUs at 4-bit, per hour | 1x MI300X, $1.85 | 1x MI300X, $1.85 |
| Revision viewed | 7999e1d33597 | 44e7d1d6caec |
| Downloads reported by the hub | 4.1M | 980.2k |
| Last observed | 2026-10-13 | 2026-10-13 |
An evaluation row appears only where at least two of these models report the same benchmark with the same stated configuration, metric, unit and setup. Different evaluators stay named in each cell. Values are shown as reported: no unit conversion, no ranking.
Other Reported Results
These results are listed for each model on its own, because the conditions needed to compare them are not stated or do not match. Two results that leave a condition blank are not assumed to share it.
jina-embeddings-v2-small-en
| Benchmark | Conditions | Result | Reported by | Revision | Date |
|---|---|---|---|---|---|
| MTEB AmazonCounterfactualClassification (en) | Configuration enTask ClassificationMetric accuracyComparison conditions not established | 71.3582 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonCounterfactualClassification (en) | Configuration enTask ClassificationMetric apComparison conditions not established | 33.9993 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonCounterfactualClassification (en) | Configuration enTask ClassificationMetric f1Comparison conditions not established | 65.3854 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonPolarityClassification | Configuration defaultTask ClassificationMetric accuracyComparison conditions not established | 82.9014 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonPolarityClassification | Configuration defaultTask ClassificationMetric apComparison conditions not established | 78.0143 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonPolarityClassification | Configuration defaultTask ClassificationMetric f1Comparison conditions not established | 82.8336 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonReviewsClassification (en) | Configuration enTask ClassificationMetric accuracyComparison conditions not established | 40.89 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB AmazonReviewsClassification (en) | Configuration enTask ClassificationMetric f1Comparison conditions not established | 39.2094 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric map_at_1Comparison conditions not established | 23.257 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric map_at_10Comparison conditions not established | 37.946 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric map_at_100Comparison conditions not established | 39.17 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric map_at_1000Comparison conditions not established | 39.181 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric map_at_3Comparison conditions not established | 32.99 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric map_at_5Comparison conditions not established | 35.468 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric mrr_at_1Comparison conditions not established | 23.542 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric mrr_at_10Comparison conditions not established | 38.057 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric mrr_at_100Comparison conditions not established | 39.289 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric mrr_at_1000Comparison conditions not established | 39.299 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric mrr_at_3Comparison conditions not established | 33.096 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric mrr_at_5Comparison conditions not established | 35.628 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric ndcg_at_1Comparison conditions not established | 23.257 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric ndcg_at_10Comparison conditions not established | 46.729 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric ndcg_at_100Comparison conditions not established | 51.901 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric ndcg_at_1000Comparison conditions not established | 52.16 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric ndcg_at_3Comparison conditions not established | 36.323 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric ndcg_at_5Comparison conditions not established | 40.767 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric precision_at_1Comparison conditions not established | 23.257 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric precision_at_10Comparison conditions not established | 7.511 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric precision_at_100Comparison conditions not established | 0.976 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric precision_at_1000Comparison conditions not established | 0.1 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric precision_at_3Comparison conditions not established | 15.339 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric precision_at_5Comparison conditions not established | 11.351 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric recall_at_1Comparison conditions not established | 23.257 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric recall_at_10Comparison conditions not established | 75.107 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric recall_at_100Comparison conditions not established | 97.582 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric recall_at_1000Comparison conditions not established | 99.573 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric recall_at_3Comparison conditions not established | 46.017 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArguAna | Configuration defaultTask RetrievalMetric recall_at_5Comparison conditions not established | 56.757 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArxivClusteringP2P | Configuration defaultTask ClusteringMetric v_measureComparison conditions not established | 44.0242 | jinaai Publisher reported |
Evaluated revision not stated | — |
| MTEB ArxivClusteringS2S | Configuration defaultTask ClusteringMetric v_measureComparison conditions not established | 35.1614 | jinaai Publisher reported |
Evaluated revision not stated | — |
SAVRN's Notes on bge-small-zh-v1.5
Twenty-four million parameters and a 0.1 GB footprint at 16-bit, rounding to zero at 8-bit and 4-bit: this is the Chinese-text embedding model you bolt onto a pipeline, not the one you rent hardware for. bge-small-zh-v1.5 draws 4.8M downloads a month, the heaviest pull among the embedding checkpoints we track, and the cheapest listed setup — one MI300X with 192 GB at $1.85 an hour — is never sized for it. It rides in spare memory on a card already serving something bigger. The 512-token window is the real constraint; longer Chinese documents get chunked before they reach the encoder.
MIT keeps the paperwork short: commercial use, modification and redistribution, provided the copyright and permission notices travel with the files. The record shows a BertModel architecture in transformers, published September 2023 and last touched October 2023, so the weights are stable enough to pin without chasing revisions. If your corpus is multilingual rather than Chinese-only, this checkpoint is the wrong shape; for Chinese-only retrieval it does the job at a footprint no host needs to price by the token.
SAVRN's Notes on jina-embeddings-v2-small-en
Turning English text into vectors is the whole job here, and the 8,192 token window lets a full contract or filing go in as one piece. It is 33 million parameters in four layers, stored in float32, about 523 MB on disk. At 16-bit it needs 0.1 GB of memory, so the cheapest Index setup, one MI300X at $1.85 an hour, is far more card than the job calls for; in practice it shares a GPU with a generator, and the publisher also ships ONNX and CoreML builds.
Apache 2.0 permits commercial use, modification and redistribution, provided you keep the license and NOTICE files and state significant changes, and access is open. Before committing, confirm an English-only model fits your corpus, and read the ALiBi paper listed against it for how the 8,192 length is reached. The training relations name the C4 backbone and the publisher's negation dataset.
Questions
Which is larger, bge-small-zh-v1.5 or jina-embeddings-v2-small-en?
jina-embeddings-v2-small-en (33M parameters) is larger than bge-small-zh-v1.5 (24M parameters), by the parameter counts their publishers report.
Which is cheaper to run, bge-small-zh-v1.5 or jina-embeddings-v2-small-en?
At 4-bit, bge-small-zh-v1.5 fits on 1x MI300X from $1.85 an hour and jina-embeddings-v2-small-en on 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists.
Can I use bge-small-zh-v1.5 commercially?
Yes. bge-small-zh-v1.5 is released under MIT License. The MIT License is a short permissive license. It permits commercial use, modification and redistribution, provided the copyright notice and permission notice are included.
Can I use jina-embeddings-v2-small-en commercially?
Yes. jina-embeddings-v2-small-en is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors.