NV-EmbedQA-E5-v5
NVIDIA's E5-family embedding model for question answering retrieval, via NIM.
Retired
NV-EmbedQA-E5-v5 appears to be discontinued or shut down — retired from NVIDIA NIM (end-of-life 2026) This page is kept as an archive record and is excluded from search engines.
About
NV-EmbedQA-E5-v5 is NVIDIA's Retrieval QA embedding model, a fine-tuned E5-Large-Unsupervised transformer encoder with 24 layers and 1,024-dimensional embeddings. It was trained on a 400K-sample blend of public QA datasets selected for commercial-friendly licensing, addressing the MS MARCO commercial-use restriction that limits many rival retrievers. NVIDIA ships it under Apache 2.0 and serves it through the NIM API with an OpenAI-compatible embeddings endpoint.
Lifecycle caveat: NVIDIA retired this model from NIM serving (end-of-life reported in 2026), superseding it with newer embedding models such as Nemotron-Embed. The weights and model card remain reference material for retrieval research. Release date is era-level (2024) since NVIDIA did not publish a precise launch date.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 119; Trending score 70/100 from researcher assessment.