The story so far
Weights, every dated article, and repo commits on one rail.
2 dated articles and 14 repo commits so far
Teal is the weights (2024-04-20), amber triangles are 2 dated articles (each links to it), grey ticks are repo commits.
Models › gte-large-v1.5 family
gte-large-en-v1.5
Alibaba-NLP Apache 2.0 out 3 years ago
434M parameters, text in, embeddings out, 8K tokens of context.
Verified 2026-10-25 Rank #38 in Embeddings · score 92.2 method v1 Sources 3 Downloads 30d 1.8M · #69 of 619 tracked
Do people use it?1.8Mdownloads, 30 days#69 of 619 we track · top 11%+4 likes, downloads +7% this week251 likesHugging Face2026-10-24
Where it's used
Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.
No API provider serves it, so its downloads are people running it themselves: 9 quantized builds and 37 Spaces.
- As a service0no API provider listed
- On own hardware9quantized builds
- Built on1313 fine-tunes · 0 adapters · 0 merges
- In apps37Hugging Face Spaces
Top Spaces running it: oceansweep/tldw ♥ 27 · rishisim/aiotsmartlabs-assistant ♥ 1 · fffffgggg54/Danbooru_0-shot_demo_vit_base_patch16_448_old ♥ 1 · fffffgggg54/Danbooru_0-shot_demo_davit_tiny_old ♥ 1
Will it fit?
Weights only, an estimate: leave room for the KV cache. Method
Fits a 16 GB card at INT4
About 0.3 GB at INT4, 0.5 GB at FP8, 1.0 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
What people built on it
22 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
22 derivatives on the Hub, most of them fine-tunes
From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.
Reference
Everything else, folded: the full tables, every source.
Benchmarks
14 rows
Benchmarks 14 rows
measured third-party leaderboard rowauthor-reported number from the model card or eval files (14)No confidence intervals unless the source gave them; settings differ between rows, so compare within a benchmark, not across. Date = the benchmark's own date when it gives one, else when we read it.
| Benchmark | Value | Setting / notes | Reported by | Date |
|---|---|---|---|---|
| author-reported LoCo · Embed text for search | 86.71 | Average (5) | model author | 2026-10-07 |
| author-reported LoCo · Embed text for search | 44.55 | QsmsumRetrieval | model author | 2026-10-07 |
| author-reported LoCo · Embed text for search | 92.61 | SummScreenRetrieval | model author | 2026-10-07 |
| author-reported LoCo · Embed text for search | 99.82 | QasperAbastractRetrieval | model author | 2026-10-07 |
| author-reported LoCo · Embed text for search | 97.81 | QasperTitleRetrieval | model author | 2026-10-07 |
| author-reported LoCo · Embed text for search | 98.74 | GovReportRetrieval | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 65.39 | Average (56) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 77.75 | Class. (12) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 47.95 | Clust. (11) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 84.63 | Pair Class. (3) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 58.5 | Reran. (4) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 57.91 | Retr. (15) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 81.43 | STS (10) | model author | 2026-10-07 |
| author-reported MTEB · Embed text for search | 30.91 | Summ. (1) | model author | 2026-10-07 |
What the card says
1 cited sentence
Summary generated, cited
The architecture is built upon a transformer++ encoder and supports a context length of up to 8192.[1]
Generated 2026-10-07 15:04 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.
Use-case evidence
2 use cases
What it is good at 2 use cases
Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.
Facts
21 fields
Listed in Embeddings
Facts
- gte-large-en-v1.5 has 434M parameters. (safetensors metadata, observed 2026-10-24)
- Context length is 8,192 tokens (8K). (config.json max_position_embeddings, observed 2026-10-24)
- License: Apache 2.0 (friction: none). Permissive; patent grant. (license file, observed 2026-10-24)
- Created 2024-04-20, last modified 2024-08-09. (HF API, observed 2026-10-24)
| Pipeline tag | sentence-similarity |
|---|---|
| Library | transformers |
| Architectures | NewModel |
| Model type | new |
| Dtype | float32 |
| Params by dtype | F32: 434,139,136 |
| Size band | S |
| MoE | no |
| Context length | 8,192 (config.json max_position_embeddings) |
| Languages | en |
| Papers | arXiv:2407.19669, arXiv:2308.03281 |
| License | Apache 2.0 (apache-2.0) |
| Hub snapshot | 2026-10-25 |
Servability
vLLM · SGLang
Servability
vLLM: not in registry · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: not in registry · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed
vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.
Hardware fit
3 estimates
Hardware fit ESTIMATE
Weights-only memory for gte-large-en-v1.5: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 1.0 GB · Edge (<4 GB) |
|---|---|
| FP8 (1 byte/param) | 0.5 GB · Edge (<4 GB) |
| INT4 (0.5 bytes/param) | 0.3 GB · Edge (<4 GB) |
Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.
All variants
22 variants
Lineage
finetune: 13quantized: 9
Top variants (22 shown, ranked by downloads)
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| locuslab/safety-classifier_gte-large-en-v1.5 | finetune | — | 435M | S | apache-2.0 | locuslab | 6 | 2.6k |
| novacardsai/gte-large-medical | finetune | — | 434M | S | — | novacardsai | 0 | 1.2k |
| cstr/gte-large-en-v1.5-GGUF | quantized | GGUF | — | — | apache-2.0 | cstr | 0 | 964 |
| mrm8488/gte-large-ft-webinstruct | finetune | — | 434M | S | — | mrm8488 | 9 | 391 |
| VibrantVista/gte-large-en-v1.5-stylejudge | finetune | — | 434M | S | — | VibrantVista | 0 | 707 |
| dpanea/skill-assignment-transformer | finetune | — | 434M | S | — | dpanea | 0 | 482 |
| Kouskousi/gte-large-en-v1.5_SEC_docs_ft_with_5_epochs | finetune | — | 434M | S | — | Kouskousi | 0 | 305 |
| lw2134/policy_gte_large_2plus | quantized | — | 434M | S | — | lw2134 | 0 | 241 |
| lw2134/policy_gte_large | quantized | — | 434M | S | — | lw2134 | 0 | 214 |
| lw2134/policy_gte_large_2 | quantized | — | 434M | S | — | lw2134 | 0 | 207 |
| openpecha/Finetuned_Alibaba_Large | finetune | — | 434M | S | — | openpecha | 0 | 159 |
| lw2134/policy_gte_large_7 | quantized | — | 434M | S | — | lw2134 | 0 | 129 |
| spl4shedEdu/gte_ESB | finetune | — | 434M | S | — | spl4shedEdu | 0 | 113 |
| Gurveer05/gte-large-eedi-2024 | finetune | — | 434M | S | — | Gurveer05 | 0 | 112 |
| ganeshanmalhotra007/model_3 | finetune | — | 434M | S | — | ganeshanmalhotra007 | 0 | 106 |
| lw2134/policy_gte_large_5 | quantized | — | 434M | S | — | lw2134 | 0 | 106 |
| spl4shedEdu/gte_ISB | finetune | — | 434M | S | — | spl4shedEdu | 0 | 92 |
| spl4shedEdu/gte_IAB | finetune | — | 434M | S | — | spl4shedEdu | 0 | 87 |
| spl4shedEdu/gte_ISM | finetune | — | 435M | S | — | spl4shedEdu | 0 | 87 |
| keisuke-miyako/gte-large-en-v1.5-onnx-int8 | quantized | int8 | — | — | — | keisuke-miyako | 0 | 0 |
| keisuke-miyako/gte-large-en-v1.5-onnx-f32 | quantized | — | — | — | — | keisuke-miyako | 0 | 0 |
| keisuke-miyako/gte-large-en-v1.5-onnx-f16 | quantized | — | — | — | — | keisuke-miyako | 0 | 0 |
Repository changelog
14 commits
Repository changelog 14 commits
| Date | Commit | SHA |
|---|---|---|
| 2024-08-09 | Update README.md | 104333d |
| 2024-08-06 | Update README.md | 4b8da4a |
| 2024-08-06 | Update README.md | 75b53a6 |
| 2024-04-26 | Update README.md | a0d6174 |
| 2024-04-22 | Add transformers.js example code (#6) | bb3242d |
| 2024-04-22 | Upload ONNX weights (#5) | 42499ad |
| 2024-04-21 | Update README.md | 791c3b7 |
| 2024-04-21 | Update README.md | 47afa7f |
| 2024-04-21 | Add sentence-transformers tag; will help with findability & MTEB filtering (#4) | a172bbf |
| 2024-04-21 | Update README.md | 0b89a03 |
| 2024-04-21 | Update README.md | a0c84dc |
| 2024-04-20 | Support sentence_transformers and fix readme (#2) | 4a1c613 |
| 2024-04-20 | add model files (#1) | b4a2b8c |
| 2024-04-20 | initial commit | 01a3a78 |
Articles and coverage
6 pieces
Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.
Analysis6
promptlayer.com276 words
GTE-Large-EN-V1.5 is a 434M parameter text embedding model from Alibaba Group's Institute for Intelligent Computing released under an Apache 2.0 license. The model supports a context length of 8192 tokens and achieves a 65.39 average score on the MTEB benchmark. It utilizes a transformer++ encoder backbone combining BERT architecture with RoPE and GLU components.written, reviewed, cited
savrn.com2,686 words
Alibaba-NLP released gte-large-en-v1.5, an open-weight sentence similarity model under Apache License 2.0 with 434M parameters and an 8,192-token context. The model requires about 1 GB of GPU memory at 16-bit precision, running on a 1x MI300X for $1.85 an hour. It achieves a 65.39 MTEB-en score and draws 1.7M downloads a month.written, reviewed, cited
- The repository ships every precision at once, totaling 6.4 GB including safetensors and ONNX copies.
- Weights are stored in float32 precision, requiring conversion to achieve the listed 16-bit memory usage.
- The model was trained on the disclosed allenai/c4 dataset under the Apache 2.0 license.
huggingpeers.co
model.aibase.com
zilliz.com2024-08-12562 words
The gte-large-en-v1.5 model achieves an MTEB Score of 65.39 and a LoCo Score of 86.71. This model utilizes a transformer++ encoder backbone and supports a maximum input of 8192 tokens.generated, cited
blog.csdn.net2025-01-09
Papers
2 papers
Papers 2
arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).
Similar models
3 models
Similar models
- LaBSE · 471M same size band (S), same task (sentence-similarity), more providers, permissive license
- e5-large-v2 · 335M same size band (S), same task (sentence-similarity), more providers, permissive license
- all-roberta-large-v1 · 355M same size band (S), same task (sentence-similarity), more providers, permissive license
Sources
3 sources
Sources
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:36 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:04 UTC · Last verified 2026-10-25
Follow gte-large-en-v1.5
Watch this model
- /feeds/models/Alibaba-NLP--gte-large-en-v1.5.xml(RSS 2.0) · every change to gte-large-en-v1.5
- /feeds/families/Alibaba-NLP--gte-large-v1.5.xml(RSS 2.0) · every change across the gte-large-v1.5 family
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.