nomic-embed-text-v2-moe

475M parameters (2 of 8 experts per token), text in, embeddings out, 2K tokens of context.

Verified 2026-10-25 Rank #42 in Embeddings · score 90.5 method v1 Sources 3 Downloads 30d 973.2k · #114 of 619 tracked

Can I run it?0.3GB at INT4Estimatefits a 16 GB cardmethod

Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24

Does it work?63.64on MTEB#17 of 64 measuredleaderboard2026-10-24

Do people use it?973.2kdownloads, 30 days#114 of 619 we track · top 18%+3 likes, downloads −14% this week512 likesHugging Face2026-10-24

Worth reading

Picked by rule from 7 tracked pieces: the lab's own post, the most shared, a field report, the latest.

Latest · 20 months ago

Nomic Embed Text V2: An Open Source, Multilingual, Moe Embedding Model

nomic.ai 2025-02-17 1 HN pts · 2 comments about the nomic family family

7 best of 7 tracked

The 2 picks above, plus these 5. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

Article[PDF] Training Sparse Mixture Of Experts Text Embedding Modelsstatic.nomic.ai · undated · official
ArticleTraining Sparse Mixture Of Experts Text Embedding Modelsstatic.nomic.ai · undated · official
ArticleNomic Embed: A Open Embedding Modelblog.nomic.ai · 2024-02-01 · 4 HN pts · official
ArticleMultimodal Embeddings for Unified Latent Space by Nomicblog.nomic.ai · 2024-06-05 · 2 HN pts · official
ArticleHivebook - The Knowledge Base for AI Agentshivebook.wiki · undated

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: 19 quantized builds and 15 Spaces.

As a service0no API provider listed
On own hardware19quantized builds
Built on2828 fine-tunes · 0 adapters · 0 merges
In apps15Hugging Face Spaces

Top Spaces running it: denniscraandijk/mtem-pruner ♥ 3 · muhammadsalmanalfaridzi/RAG-GutHib ♥ 1 · gweakliem/embed_experiments ♥ 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, 0.6 GB at FP8, 1.1 GB at BF16: weights only, so leave room for the KV cache. Every expert must be resident, so total parameters count.

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.

#17 of 64 on MTEB so far

63.64. Each tick is a model on this site with a measured MTEB row; author-reported numbers stay in Reference. Mean(task) x100 over 41 tasks. 41 tasks, split=test

Source · leaderboard · 2026-10-24

The story so far

Weights, every dated article, and repo commits on one rail.

Most coverage came in the first two weeks

Teal is the weights (2025-02-07), amber triangles are 3 dated articles (each links to it), grey ticks are repo commits.

What people built on it

47 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

47 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.

GGUF nomic-ai/nomic-embed-text-v2-moe-GGUF for llama.cpp, Ollama, LM Studio 259.8k downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

Benchmarks

5 rows

Benchmarks 5 rows

measured third-party leaderboard row (3)author-reported number from the model card or eval files (2)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
measured MTEB · Embed text for search	54.81 nDCG@10 x100 (mean of 10 tasks)	10 tasks, split=test	leaderboard	2026-10-24
measured MTEB · Embed text for search	63.64 mean(task) x100 over 41 tasks	41 tasks, split=test	leaderboard	2026-10-24
measured MTEB Reranking · Rerank search results	45.23 MAP x100 (mean of 2 tasks)	2 tasks, split=test	leaderboard	2026-10-24
author-reported BEIR · Embed text for search	52.86	Nomic Embed v2	model author	2026-10-07
author-reported MIRACL · Embed text for search	65.8	Nomic Embed v2	model author	2026-10-07

What the card says

2 cited sentences

Summary generated, cited

nomic-embed-text-v2-moe is a multilingual mixture of experts text embedding model with 475M total parameters.^[1]^[2] The architecture utilizes 8 experts and supports flexible embedding dimensions from 768 to 256.^[3]^[4]

parameters human
“Mixture of Experts (MoE) architecture”
num experts
“Supports flexible dimension from 768 to 256”

Generated 2026-10-07 15:03 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

3 use cases

What it is good at 3 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

24 fields

Listed in Embeddings · MoE

Facts

nomic-embed-text-v2-moe has 475M parameters. (safetensors metadata, observed 2026-10-24)
nomic-embed-text-v2-moe is a mixture-of-experts model with 8 experts, routing 2 per token. (config.json, observed 2026-10-24)
Context length is 2,048 tokens (2K). (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 2025-02-07, last modified 2025-04-01. (HF API, observed 2026-10-24)

Pipeline tag	sentence-similarity
Library	sentence-transformers
Architectures	NomicBertModel
Model type	nomic_bert
Dtype	float32
Params by dtype	F32: 475,292,928
Size band	S
MoE	yes
Experts	8
Experts per token	2
Context length	2,048 (config.json max_position_embeddings)
Languages	en, es, fr, de, it, pt, pl, nl, tr, ja, vi, ru, id, ar, cs, ro, sv, el, uk, zh, hu, da, no, hi, fi, bg, ko, sk, th, he, ca, lt, fa, ms, sl, lv, mr, bn, sq, cy, be, ml, kn, mk, ur, fy, te, eu, sw, so, sd, uz, co, hr, gu, ce, eo, jv, la, zu, mn, si, ga, ky, tg, my, km, mg, pa, sn, ha, ht, su, gd, ny, ps, ku, am, ig, lo, mi, nn, sm, yi, st, tl, xh, yo, af, ta, tn, ug, az, ba, bs, dv, et, gl, gn, gv, hy
Papers	arXiv:2502.07972, arXiv:2205.13147
License	Apache 2.0 (apache-2.0)
Hub snapshot	2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · NomicBertModel · 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 nomic-embed-text-v2-moe: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)	1.1 GB · Edge (<4 GB)
FP8 (1 byte/param)	0.6 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. MoE: all experts must be resident, so total parameters count.

All variants

40 variants

Lineage

Fine-tuned from nomic-ai/nomic-embed-text-v2-moe-unsupervised (same organization) Family: nomic-ai/nomic 47 derivatives on the Hub

Base model: nomic-ai/nomic-embed-text-v2-moe-unsupervised

finetune: 28quantized: 19

Top variants (40 shown, ranked by downloads)

Variant	Relation	Quant	Params	Band	License	Author	Likes	Downloads
cnmoro/nomic-embed-text-v2-moe-distilled-high-quality	finetune	—	192M	S	apache-2.0	cnmoro	5	408.2k
nomic-ai/nomic-embed-text-v2-moe-GGUF	quantized	GGUF	—	—	apache-2.0	nomic-ai	80	259.8k
basilisk78/nomic-v2-tuned-1	finetune	—	475M	S	apache-2.0	basilisk78	1	168k
cnmoro/static-nomic-eng-ptbr-tiny	finetune	—	12.5M	S	apache-2.0	cnmoro	4	78k
ggml-org/Nomic-Embed-Text-V2-GGUF	quantized	GGUF	—	—	apache-2.0	ggml-org	18	45k
cstr/nomic-embed-text-v2-moe-GGUF	quantized	GGUF	—	—	apache-2.0	cstr	1	3.7k
indiansatoshi/nomic-embed-text-v2-moe-GGUF	quantized	GGUF	—	—	apache-2.0	indiansatoshi	2	1.7k
olegshulyakov/nomic-embed-text-v2-moe-GGUF	quantized	GGUF	—	—	apache-2.0	olegshulyakov	1	1.7k
cnmoro/static-nomic-384-pten	finetune	—	12.8M	S	apache-2.0	cnmoro	2	1.5k
aviallon/nomic-ai-nomic-embed-text-v2-moe	quantized	GGUF	—	—	apache-2.0	aviallon	0	1.6k
Abdelkareem/nomic-embed-text-v2-moe_distilled	finetune	—	64.0M	S	mit	Abdelkareem	0	998
cnmoro/static-nomic-384-pten-v2-st	finetune	—	106M	S	apache-2.0	cnmoro	2	892
benvezzani/nomic-embed-text-v2-moe-Q4_K_M-GGUF	quantized	GGUF	—	—	apache-2.0	benvezzani	2	740
keisuke-miyako/nomic-embed-text-v2-moe-gguf-q8_0	quantized	gguf	—	—	—	keisuke-miyako	1	785
YeonwooSung/nomic-embed-text-v2-moe-msmarco-bpr	finetune	—	475M	S	—	YeonwooSung	1	669
cnmoro/static-nomic-384-pten-v2	finetune	—	12.8M	S	apache-2.0	cnmoro	2	533
eyassa/nomic-embed-text-v2-moe-Q8_0-GGUF	quantized	Q8_0	—	—	apache-2.0	eyassa	0	631
wp-airik/nomic-embed-text-v2-moe-Q8_0-GGUF	quantized	Q8_0	—	—	apache-2.0	wp-airik	0	550
benvezzani/nomic-embed-text-v2-moe-Q8_0-GGUF	quantized	Q8_0	—	—	apache-2.0	benvezzani	0	504
teoyidu/nomic-moe-turkish-v1	finetune	—	—	—	apache-2.0	teoyidu	1	286
AbteeXAILab/lumynax-embed-nomic-v2-moe	finetune	—	475M	S	apache-2.0	AbteeXAILab	0	335
cnmoro/portuguese-nomic-embed-text-v2-moe	quantized	—	311M	S	unknown	cnmoro	1	284
Cactus-Compute/nomic-embed-text-v2-moe	finetune	—	—	—	apache-2.0	Cactus-Compute	1	268
majentik/nomic-embed-text-v2-moe-GGUF-Q8_0	quantized	GGUF	—	—	apache-2.0	majentik	0	309
pyarn/nomic-embed-text-v2-moe-Q8_0-GGUF	quantized	Q8_0	—	—	apache-2.0	pyarn	0	298
cnmoro/static-nomic-eng-ptbr	finetune	—	128M	S	apache-2.0	cnmoro	3	144
IoannisKat1/nomic-embed-text-v2-moe-matryoshka1	finetune	—	475M	S	apache-2.0	IoannisKat1	0	293
tsss1/modernbert-embed-base-legal-matryoshka-2	finetune	—	475M	S	apache-2.0	tsss1	0	249
Torkovweryyy/nomic-embed-text-v2-moe-Q8_0-GGUF	quantized	Q8_0	—	—	apache-2.0	Torkovweryyy	0	234
majentik/nomic-embed-text-v2-moe-GGUF-IQ4_XS	quantized	GGUF	—	—	apache-2.0	majentik	0	233
majentik/nomic-embed-text-v2-moe-GGUF-Q5_K_M	quantized	GGUF	—	—	apache-2.0	majentik	0	231
majentik/nomic-embed-text-v2-moe-GGUF-Q4_K_M	quantized	GGUF	—	—	apache-2.0	majentik	0	227
tsss1/expressvpn_embeddingmodel	finetune	—	475M	S	apache-2.0	tsss1	0	223
Abdelkareem/khatib-v0.1	finetune	—	475M	S	—	Abdelkareem	0	212
DGSMsRzJ6xC2JthtHG9W/nomic-v2-tuned-1	finetune	—	475M	S	mit	DGSMsRzJ6xC2JthtHG9W	1	131
DGSMsRzJ6xC2JthtHG9W/nomic-v2-tuned-1.4	finetune	—	475M	S	apache-2.0	DGSMsRzJ6xC2JthtHG9W	0	173
Demircan12/nomic-embed-text-v2-moe-YeditepeFT	finetune	—	475M	S	—	Demircan12	0	173
netrias/nomic-embed-text-v2-moe-onnx	quantized	—	—	—	apache-2.0	netrias	0	167
DGSMsRzJ6xC2JthtHG9W/nomic-v2-tuned-1.2	finetune	—	475M	S	apache-2.0	DGSMsRzJ6xC2JthtHG9W	0	150
cnmoro/static-nomic-distilled-vocab-quantized	finetune	-quant	1.78M	S	mit	cnmoro	0	123

Repository changelog

30 commits

Repository changelog 30 commits

Date	Commit	SHA
2025-04-01	Update README.md	1066b65
2025-04-01	Update README.md	e89d1c9
2025-03-31	Update README.md (#15)	302c115
2025-03-11	fix: add paper order	45301cc
2025-03-11	fix: add paper	d0f89cf
2025-02-15	Update config_sentence_transformers.json	f6a8873
2025-02-15	Update config.json	11920f9
2025-02-14	Update config_sentence_transformers.json	13dae9b
2025-02-14	Update config_sentence_transformers.json	c38f2cb
2025-02-14	update prompts per task	4d3c2ef
2025-02-13	Update README.md	1bc19bc
2025-02-12	Update README.md (#4)	747a5f0
2025-02-12	Update README.md	94bdb17
2025-02-12	Update README.md	dbb2e05
2025-02-12	Extra typo, and add more focus on Nomic Embed v2 in the table (#3)	2eb9b8b
2025-02-12	Update model card slightly (#2)	3e60adb
2025-02-12	Update README.md	48681e7
2025-02-11	Update README.md	93b80e7
2025-02-11	Update README.md	ac3a5fe
2025-02-11	Update README.md	3b160e9
2025-02-11	Update README.md	24132aa
2025-02-11	Update README.md	b2554b1
2025-02-11	Update README.md	fcfe1ed
2025-02-11	Update README.md	a69ccd7
2025-02-11	Update config_sentence_transformers.json	c5fb980
2025-02-11	Upload model	c8bdf8c
2025-02-11	Add new SentenceTransformer model	e650938
2025-02-10	Upload tokenizer	8e10993
2025-02-07	Upload model	f0bcc88
2025-02-07	initial commit	c5783e0

Articles and coverage

7 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.

Official5

[PDF] Training Sparse Mixture Of Experts Text Embedding Models

static.nomic.aiofficialabout the nomic family6,863 words

Nomic Embed v2 is the first general-purpose sparse Mixture of Experts text embedding model. The 305M parameter model uses a 768 embedding dimension and achieves a BEIR score of 52.86 and a MIRACL score of 65.80. It outperforms same-sized models and matches performance of models twice its size, with code available at github.com/nomic-ai/contrastors.written, reviewed, cited

The model achieves competitive performance with models twice its size while using fewer active parameters.
Multilingual models typically require 3-5x more parameters than monolingual models to achieve comparable English performance.
The router dynamically assigns each input token to a subset of experts using Top-K routing based on highest probabilities.

Training Sparse Mixture Of Experts Text Embedding Models

static.nomic.aiofficialabout the nomic family6,861 words

Nomic Embed v2 is the first general-purpose Mixture of Experts text embedding model, activating only a subset of parameters to reduce inference latency and memory usage. The 305M parameter model achieves a MIRACL score of 65.80 with a 768 embedding dimension, outperforming peers in its class while maintaining competitive performance against models twice its size. Training requires an auxiliary load balancing loss to prevent expert collapse, and the model supports both monolingual and multilingual benchmarks.written, reviewed, cited

The model uses Top-K routing to assign each token to the top-k experts with the highest probabilities.

Nomic Embed: A Open Embedding Model

blog.nomic.ai2024-02-01officialabout the nomic familyHN 4 pts · 0 comments

Nomic Embed Text V2: An Open Source, Multilingual, Moe Embedding Model

nomic.ai2025-02-17officialabout the nomic familyHN 1 pts · 2 comments

Multimodal Embeddings for Unified Latent Space by Nomic

blog.nomic.ai2024-06-05officialabout the nomic familyHN 2 pts · 0 comments

Analysis2

nomic-embed-text-v2-moe - Ollama

ollama.com327 words

Hivebook - The Knowledge Base for AI Agents

hivebook.wikiabout the nomic family1,504 words

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).

Training Sparse Mixture Of Experts Text Embedding Models this model

Zach Nussbaum, Brandon Duderstadt · 2025-02-11 · PDF · HF paper page (12 upvotes)
Matryoshka Representation Learning cited in card

Aditya Kusupati, Gantavya Bhatt, Aniket Rege +8 · 2022-05-26 · PDF · HF paper page (30 upvotes)

Similar models

3 models

Similar models

LaBSE · 471M
same size band (S), same task (sentence-similarity), more providers, permissive license
gte-large-en-v1.5 · 434M
same size band (S), same task (sentence-similarity), permissive license
snowflake-arctic-embed-l-v2.0 · 568M
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 21:35 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:03 UTC · Last verified 2026-10-25

Follow nomic-embed-text-v2-moe

Watch this model

/feeds/models/nomic-ai--nomic-embed-text-v2-moe.xml(RSS 2.0) · every change to nomic-embed-text-v2-moe
/feeds/families/nomic-ai--nomic.xml(RSS 2.0) · every change across the nomic family

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.