The story so far
Weights, every dated article, and repo commits on one rail.
1 dated article and 10 repo commits so far
Teal is the weights (2024-02-08), amber triangles are 1 dated article (each links to it), grey ticks are repo commits.
multilingual-e5-large-instruct
560M parameters, text in, embeddings out, 514 tokens of context. 2 providers serve it from $0.010 per million input tokens.
Verified 2026-10-25 Rank #15 in Embeddings · score 101.5 method v1 Sources 3 Downloads 30d 1.7M · #74 of 619 tracked
Changed this week: 1 new quantized build on the Hub (265 in all).2026-10-24
Do people use it?1.7Mdownloads, 30 days#74 of 619 we track · top 12%+4 likes, downloads +4% this week655 likesHugging Face2026-10-24
Where to run it
Live hosted endpoints, cheapest first, USD per million tokens. Ties break on output price, then provider name.
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.
Served by 2 API providers; heavily run locally (265 quantized builds); a big base for builders (175 fine-tunes, adapters and merges).
- As a service2DeepInfra, HF Inference
- On own hardware265quantized builds
- Built on175170 fine-tunes · 5 adapters · 0 merges
- In apps100+Hugging Face Spaces
Top Spaces running it: yourbench/advanced ♥ 45 · lfoppiano/document-qa ♥ 16 · perctrix/AI-Lawyer ♥ 8 · atlasia/moul_lmemes ♥ 8
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, 0.7 GB at FP8, 1.3 GB at BF16: weights only, so leave room for the KV cache.
Estimate params × bytes per param × 1.2 · how we estimate
How it scores
Third-party measurements only; author-reported numbers stay in Reference, marked as such.
#3 of 33 on MTEB Reranking so far
48.74. Each tick is a model on this site with a measured MTEB Reranking row; author-reported numbers stay in Reference. MAP x100 (mean of 2 tasks). 2 tasks, split=test
Source · leaderboard · 2026-10-24
What people built on it
440 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.
440 derivatives on the Hub, most of them quantized
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
3 rows
Benchmarks 3 rows
measured third-party leaderboard row (3)author-reported number from the model card or eval filesNo 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 |
|---|---|---|---|---|
| measured MTEB · Embed text for search | 53.47 nDCG@10 x100 (mean of 10 tasks) | 10 tasks, split=test | leaderboard | 2026-10-24 |
| measured MTEB · Embed text for search | 65.56 mean(task) x100 over 41 tasks | 41 tasks, split=test | leaderboard | 2026-10-24 |
| measured MTEB Reranking · Rerank search results | 48.74 MAP x100 (mean of 2 tasks) | 2 tasks, split=test | leaderboard | 2026-10-24 |
What the card says
2 cited sentences
Summary generated, cited
The architecture consists of 24 layers and an embedding size of 1024.[1] Released under the MIT license, this multilingual model supports 100 languages.[2][3]
- “This model has 24 layers and the embedding size is 1024.”
- license
- “It supports 100 languages from xlm-roberta”
Generated 2026-10-07 15:01 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.
Use-case evidence
4 use cases
What it is good at 4 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.
- Embed text for searchmeasuredMTEB(eng, v2) retrieval 53.47MTEB(eng, v2) 65.56decision-model estimate p=1.00+4 more
- Work across languagesstructuraldecision-model estimate p=1.00name: multilingualtag: multilingual+1 more
- Run on a laptop or phonestructuraldecision-model estimate p=0.76fits: edge
- Chat and follow instructionsstructuralname: instruct
Facts
21 fields
Listed in Embeddings
Facts
- multilingual-e5-large-instruct has 560M parameters. (safetensors metadata, observed 2026-10-24)
- Context length is 514 tokens (514). (config.json max_position_embeddings, observed 2026-10-24)
- License: MIT (friction: none). Permissive. (license file, observed 2026-10-24)
- Created 2024-02-08, last modified 2025-07-10. (HF API, observed 2026-10-24)
| Pipeline tag | feature-extraction |
|---|---|
| Library | sentence-transformers |
| Architectures | XLMRobertaModel |
| Model type | xlm-roberta |
| Dtype | float16 |
| Params by dtype | F16: 559,890,432 |
| Size band | S |
| MoE | no |
| Context length | 514 (config.json max_position_embeddings) |
| Languages | multilingual, af, am, ar, as, az, be, bg, bn, br, bs, ca, cs, cy, da, de, el, en, eo, es, et, eu, fa, fi, fr, fy, ga, gd, gl, gu, ha, he, hi, hr, hu, hy, id, is, it, ja, jv, ka, kk, km, kn, ko, ku, ky, la, lo, lt, lv, mg, mk, ml, mn, mr, ms, my, ne, nl, no, om, or, pa, pl, ps, pt, ro, ru, sa, sd, si, sk, sl, so, sq, sr, su, sv, sw, ta, te, th, tl, tr, ug, uk, ur, uz, vi, xh, yi, zh |
| Papers | arXiv:2402.05672, arXiv:2401.00368, arXiv:2104.08663, arXiv:2210.07316 |
| License | MIT (mit) |
| Hub snapshot | 2026-10-25 |
Servability
vLLM · SGLang
Servability
vLLM: supported · XLMRobertaModel · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · XLMRobertaModel · 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 multilingual-e5-large-instruct: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.
| BF16 (2 bytes/param) | 1.3 GB · Edge (<4 GB) |
|---|---|
| FP8 (1 byte/param) | 0.7 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
40 variants
Lineage
finetune: 170quantized: 265adapter: 5
Top variants (40 shown, ranked by downloads)
| Variant | Relation | Quant | Params | Band | License | Author | Likes | Downloads |
|---|---|---|---|---|---|---|---|---|
| ytu-ce-cosmos/turkish-e5-large | finetune | — | 560M | S | mit | ytu-ce-cosmos | 35 | 97.6k |
| OpenMed/OpenMed-PII-mClinicalE5-Large-560M-v1 | finetune | — | 559M | S | apache-2.0 | OpenMed | 0 | 84.6k |
| OpenMed/OpenMed-PII-German-mClinicalE5-Large-560M-v1 | finetune | — | 559M | S | apache-2.0 | OpenMed | 0 | 19.5k |
| McGill-NLP/AfriE5-Large-instruct | finetune | — | 560M | S | mit | McGill-NLP | 1 | 12.7k |
| Sami92/multiling-e5-large-instruct-claim-matching | finetune | — | 560M | S | cc-by-4.0 | Sami92 | 0 | 4.1k |
| mjwong/multilingual-e5-large-instruct-xnli-anli | finetune | — | 560M | S | cc-by-nc-4.0 | mjwong | 1 | 3.3k |
| Ludo33/e5_General_VIDE_10282025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 2k |
| Lauther/measuring-embeddings-v3 | finetune | — | 560M | S | — | Lauther | 0 | 1.4k |
| Ludo33/e5_General_09092025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 1.3k |
| Ludo33/e5_General_13082025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 1.2k |
| mjwong/multilingual-e5-large-instruct-xnli | finetune | — | 560M | S | mit | mjwong | 0 | 1.2k |
| Impulse2000/multilingual-e5-large-instruct-GGUF | quantized | GGUF | — | — | mit | Impulse2000 | 1 | 991 |
| Ludo33/e5_RSE_v2 | finetune | — | 560M | S | mit | Ludo33 | 0 | 1k |
| Ludo33/e5_EC_v2 | finetune | — | 560M | S | mit | Ludo33 | 0 | 1k |
| Ludo33/e5_Biodiversite_v2 | finetune | — | 560M | S | mit | Ludo33 | 0 | 1k |
| Ludo33/e5_Dechets_v2 | finetune | — | 560M | S | mit | Ludo33 | 0 | 1k |
| Ludo33/e5_Sentiment_General_v2 | finetune | — | 560M | S | mit | Ludo33 | 0 | 979 |
| Lauther/measuring-embeddings-v4.2 | quantized | — | 560M | S | — | Lauther | 1 | 882 |
| Ludo33/e5_Eau_v4 | finetune | — | 560M | S | mit | Ludo33 | 0 | 893 |
| Ludo33/e5_EC_MultiLabel_08092025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 863 |
| Ludo33/e5_Mobilite_MultiLabel_08092025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 848 |
| Ludo33/e5_Eau_MultiLabel_08092025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 830 |
| denniscraandijk/dutch-english-multilingual-e5-large-instruct | finetune | — | 352M | S | mit | denniscraandijk | 0 | 825 |
| jester6136/multilingual-e5-large-instruct-m2v | finetune | — | 64.0M | S | mit | jester6136 | 2 | 578 |
| Ludo33/e5_EC_MultiLabel_12082025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 678 |
| Ludo33/e5_Biodiversite_MultiLabel_06082025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 666 |
| yoeven/multilingual-e5-large-instruct-Q5_0-GGUF | quantized | Q5_0 | — | — | mit | yoeven | 2 | 487 |
| Wasserstoff-AI/Legal-Embed-intfloat-multilingual-e5-large-instruct | finetune | — | 560M | S | apache-2.0 | Wasserstoff-AI | 1 | 533 |
| Content-AI/multilingual-e5-large-instruct-Q8_0-GGUF | quantized | Q8_0 | — | — | mit | Content-AI | 0 | 576 |
| FareedKhan/intfloat_multilingual-e5-large-instruct_FareedKhan_prime_synthetic_data_2k_2_4 | finetune | — | 560M | S | — | FareedKhan | 0 | 554 |
| Data-Lab/multilingual-e5-large-instruct-embedder-tgd | finetune | — | 560M | S | — | Data-Lab | 0 | 521 |
| hamtaai/e5-large-instruct-hadith | finetune | — | 560M | S | apache-2.0 | hamtaai | 0 | 521 |
| yoeven/multilingual-e5-large-instruct-Q5_K_M-GGUF | quantized | GGUF | — | — | mit | yoeven | 2 | 419 |
| Ludo33/e5_Eau_MultiLabel_11082025 | finetune | — | 560M | S | mit | Ludo33 | 0 | 483 |
| Gomez12/multilingual-e5-large-instruct-Q8_0-GGUF | quantized | Q8_0 | — | — | mit | Gomez12 | 1 | 432 |
| Data-Lab/multilingual-e5-large-instruct-embedder_distill-tgd | finetune | — | 560M | S | — | Data-Lab | 0 | 465 |
| Zenabius/multilingual-e5-large-instruct-Q8_0-GGUF | quantized | Q8_0 | — | — | mit | Zenabius | 0 | 459 |
| Gameselo/french-multilingual-e5-large-instruct | quantized | — | 342M | S | mit | Gameselo | 0 | 454 |
| alphaedge-ai/multilingual-e5-large-instruct-spa-16384 | quantized | — | 321M | S | mit | alphaedge-ai | 0 | 452 |
| jacrecra/dutch-english-multilingual-e5-large-instruct | quantized | — | 352M | S | mit | jacrecra | 0 | 449 |
Repository changelog
10 commits
Repository changelog 10 commits
| Date | Commit | SHA |
|---|---|---|
| 2025-07-10 | Update README.md | 274baa4 |
| 2025-02-17 | Ensuring compatibility with the sentence-transformers library (#24) | 84344a2 |
| 2025-02-17 | Update instruction with infinity (#23) | 1a50ec2 |
| 2024-09-26 | Adding ONNX file of this model (#17) | c9e87c7 |
| 2024-02-15 | Update README.md | baa7be4 |
| 2024-02-15 | Add Sentence Transformers support (#2) | c406bc6 |
| 2024-02-14 | Add link to new paper (#1) | 2696861 |
| 2024-02-08 | update README.md | b9fbfc1 |
| 2024-02-08 | upload model weights | d9f1d8f |
| 2024-02-08 | initial commit | 5900853 |
All providers
2 providers · 2 endpoints
Where to run 2 providers · 2 endpoints
Hosted endpoints observed for multilingual-e5-large-instruct, sorted by input price (USD per million tokens). Each row names its observer and date; prices change daily, so check the source before quoting. Cheapest observed input price: $0.01/M.
| Provider | Model ref | Context | Quant | $/M in | $/M out | tok/s | TTFT | Tools | Status | Observed |
|---|---|---|---|---|---|---|---|---|---|---|
| DeepInfra hf, deepinfra | intfloat/multilingual-e5-large-instruct | 512 | — | $0.01 | — | — | — | — | live | 2026-10-25 |
| HF Inference hf | intfloat/multilingual-e5-large-instruct | — | — | — | — | — | — | — | live | 2026-10-25 |
Articles and coverage
2 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.
Analysis2
together.aiauthority71 words
superlinked.com2026-08-04about the e5 large family727 words
The intfloat E5 family provides MIT-licensed embedding models including e5-base-v2 and multilingual-e5-large, all of which truncate inputs at 512 tokens. The multilingual-e5-large-instruct variant costs $0.0076 per million tokens and utilizes an XLM-RoBERTa-large backbone covering roughly 100 languages. Serving speeds reach 53,200 tokens/sec for base-v2 on a single L4 GPU, but omitting required role prefixes causes severe performance degradation.written, reviewed, cited
- The e5-base-v2 parameter count is not stated on its model card, though 109M is the standard BERT-base figure.
- The instruct variant replaces the prefix scheme by using a one-sentence task instruction format for queries.
Papers
4 papers
Papers 4
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
- bge-reranker-large · 560M same size band (S), same task (feature-extraction), fewer providers, permissive license
- Qwen3-Embedding-0.6B · 596M same size band (S), same task (feature-extraction), fewer providers, permissive license
- jina-embeddings-v3 · 572M same size band (S), same task (feature-extraction)
Sources
3 sources
Sources
Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:33 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:01 UTC · Last verified 2026-10-25
Follow multilingual-e5-large-instruct
Watch this model
- /feeds/models/intfloat--multilingual-e5-large-instruct.xml(RSS 2.0) · every change to multilingual-e5-large-instruct
- /feeds/families/intfloat--e5-large.xml(RSS 2.0) · every change across the e5-large family
Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.