Skip to main content
← Back to news

NVIDIA releases llama-nv-embed-reasoning-3b: 3.2B embeddings with reasoning

Model releases
Source:Hugging Face (NVIDIA)
A bidirectional 3.2B Embedding model derived from Llama 3.2 3B: it adds a reasoning step before computing the vector, aimed at Retrieval and semantic search that also run on 16GB GPUs.
Read original article

Related articles

RAG: fai domande ai tuoi documenti

RAG, Retrieval Augmented Generation, è il modo più semplice per far rispondere un modello locale a domande sui tuoi documenti: embedding, database vettoriale e ciclo di recupero, con un esempio concreto.

Get the next local-AI briefing

Once a week, the top selection from iaLocale: articles, news and resources on local AI, and more.

After signing up you receive a confirmation email (double opt-in). Without confirmation we send nothing. No spam, we promise.

Related resources

Tools and resources from the iaLocale archive relevant to this content.

🖥️

NVIDIA RTX 4060 Ti 16GB

Hardware

16GB VRAM at an entry-level price: the right choice for quantized 13-14B models. Good headroom for personal use.

🖥️

NVIDIA RTX 4070 Ti Super 16GB

Hardware

16GB VRAM with 672 GB/s bandwidth: high-end performance at a contained price. Among the best VRAM/price ratios.

🎮

NVIDIA RTX 5060 Ti 16GB

Hardware

Il 16GB d'ingresso della generazione Blackwell: GB206, 16GB GDDR7 a 448 GB/s, 180W, PCIe 5.0 x8. Lancio a 429 USD: il modo più economico di far girare Qwen3.8-27B e i modelli 30B+ in Q4 su una singola scheda. Prezzo street 2026 più alto per la carenza di GDDR7.

🎮

NVIDIA RTX 5070 Ti 16GB

Hardware

Blackwell GB203 da fascia alta: 16GB GDDR7 su bus 256-bit (896 GB/s), 8.960 CUDA core, 300W. Il 16GB rende possibile Qwen3.8-27B o i modelli 30B+ in Q4 con contesto ampio. Lancio a 749 USD, febbraio 2025.

Browse all resources