feat: add nomic-embed-text-v2-moe via Candle backend

Cargo.toml

@@ -52,13 +52,14 @@ ort-download-binaries-rustls-tls = ["ort/download-binaries", "ort/tls-rustls"]
 ort-load-dynamic = ["ort/load-dynamic"]
 
 qwen3 = ["dep:candle-core", "dep:candle-nn", "hf-hub"]
+nomic-v2-moe = ["dep:candle-core", "dep:candle-nn", "hf-hub"]
 
 image-models = ["image"]
-mkl = ["qwen3","dep:intel-mkl-src", "candle-nn/mkl", "candle-core/mkl"]
-accelerate = ["qwen3", "dep:accelerate-src", "candle-core/accelerate", "candle-nn/accelerate"]
-cuda = ["qwen3", "candle-core/cuda", "candle-nn/cuda"]
-cudnn = ["qwen3", "candle-core/cudnn", "candle-nn/cudnn", "cuda"]
-metal = ["qwen3", "candle-core/metal", "candle-nn/metal"]
+mkl = ["qwen3", "nomic-v2-moe", "dep:intel-mkl-src", "candle-nn/mkl", "candle-core/mkl"]
+accelerate = ["qwen3", "nomic-v2-moe", "dep:accelerate-src", "candle-core/accelerate", "candle-nn/accelerate"]
+cuda = ["qwen3", "nomic-v2-moe", "candle-core/cuda", "candle-nn/cuda"]
+cudnn = ["qwen3", "nomic-v2-moe", "candle-core/cudnn", "candle-nn/cudnn", "cuda"]
+metal = ["qwen3", "nomic-v2-moe", "candle-core/metal", "candle-nn/metal"]
 
 # This feature does not change any code, but is used to limit tests if
 # the user does not have `optimum-cli` or even python installed.

README.md

@@ -46,6 +46,7 @@
 - [**jinaai/jina-embeddings-v2-base-code**](https://huggingface.co/jinaai/jina-embeddings-v2-base-code)
 - [**jinaai/jina-embeddings-v2-base-en**](https://huggingface.co/jinaai/jina-embeddings-v2-base-en)
 - [**google/embeddinggemma-300m**](https://huggingface.co/google/embeddinggemma-300m)
+- [**nomic-ai/nomic-embed-text-v2-moe**](https://huggingface.co/nomic-ai/nomic-embed-text-v2-moe) - requires `nomic-v2-moe` feature (candle backend)
 - [**Qwen/Qwen3-Embedding-0.6B**](https://huggingface.co/Qwen/Qwen3-Embedding-0.6B) - requires `qwen3` feature (candle backend)
 - [**Qwen/Qwen3-Embedding-4B**](https://huggingface.co/Qwen/Qwen3-Embedding-4B) - requires `qwen3` feature (candle backend)
 - [**Qwen/Qwen3-Embedding-8B**](https://huggingface.co/Qwen/Qwen3-Embedding-8B) - requires `qwen3` feature (candle backend)
@@ -151,6 +152,31 @@ let embeddings = model.embed(&["query: ...", "passage: ..."])?;
 println!("Embeddings length: {}", embeddings.len());
 ```
 
+### Nomic Embed Text v2 MoE
+
+The [nomic-embed-text-v2-moe](https://huggingface.co/nomic-ai/nomic-embed-text-v2-moe) model is available behind the `nomic-v2-moe` feature flag (candle backend). First general-purpose MoE embedding model with 100+ language support.
+
+```toml
+[dependencies]
+fastembed = { version = "5", features = ["nomic-v2-moe"] }
+```
+
+```rust
+use candle_core::{DType, Device};
+use fastembed::NomicV2MoeTextEmbedding;
+
+let device = Device::Cpu;
+let model = NomicV2MoeTextEmbedding::from_hf(
+    "nomic-ai/nomic-embed-text-v2-moe",
+    &device,
+    DType::F32,
+    512,
+)?;
+
+let embeddings = model.embed(&["search_query: ...", "search_document: ..."])?;
+println!("Embeddings length: {}", embeddings.len());
+```
+
 ### Sparse Text Embeddings
 
 ```rust

src/lib.rs

@@ -109,3 +109,7 @@ pub use crate::reranking::{
 // For Qwen3 (candle backend)
 #[cfg(feature = "qwen3")]
 pub use crate::models::qwen3::{Config as Qwen3Config, Qwen3TextEmbedding};
+
+// For Nomic Embed Text v2 MoE (candle backend)
+#[cfg(feature = "nomic-v2-moe")]
+pub use crate::models::nomic_v2_moe::{NomicConfig, NomicV2MoeTextEmbedding};

src/models/mod.rs

@@ -10,6 +10,9 @@ pub mod text_embedding;
 #[cfg(feature = "qwen3")]
 pub mod qwen3;
 
+#[cfg(feature = "nomic-v2-moe")]
+pub mod nomic_v2_moe;
+
 pub trait ModelTrait {
     type Model;
     fn get_model_info(model: &Self::Model) -> Option<&ModelInfo<Self::Model>>;