We use cookies. Cookies and browser storage keep you signed in and remember your preferences. With your permission we also measure how the site is used. You can any time. See our cookie policy.

New here? See how it works A 1-minute tour of the playground, models and your API key.

NVIDIA: Nemotron 3.5 Embed 8B (free) vs Sentence Transformers: all-mpnet-base-v2

A side-by-side comparison of NVIDIA: Nemotron 3.5 Embed 8B (free) (Free) and Sentence Transformers: all-mpnet-base-v2 ($0.02 / 1M tokens) using current USD prices and capabilities from the api14 catalog. Both run through the same OpenAI-compatible API.

Specifications

	NVIDIA: Nemotron 3.5 Embed 8B (free)	Sentence Transformers: all-mpnet-base-v2
Provider	NVIDIA	Sentence Transformers
Input	-	$0.02 / 1M tokens
Context window	32,768	512
Max output	29,491	460
Input	text, image	text
Output	embeddings	embeddings
Tool calling
Structured output
Reasoning control
Added	Oct 21, 2026	Nov 17, 2025

Key differences

NVIDIA: Nemotron 3.5 Embed 8B (free) reads 64.0× more context (32,768 tokens).

In depth

NVIDIA: Nemotron 3.5 Embed 8B (free) is an embedding model from NVIDIA, listed under the model id nvidia/nemotron-3.5-embed-8b:free. Sentence Transformers: all-mpnet-base-v2 is an embedding model from Sentence Transformers, listed under the model id sentence-transformers/all-mpnet-base-v2. Both belong to the embeddings category, so neither one generates chat completions or free text. Each takes an input and returns a vector, which you then use for search, clustering, deduplication, classification or similar downstream work.

The clearest difference between them is context length. Nemotron 3.5 Embed 8B accepts up to 32768 tokens, while all-mpnet-base-v2 accepts up to 512 tokens. That is a large gap, and it changes what you can feed each model in a single call. Nemotron 3.5 Embed 8B also accepts both text and image input, whereas all-mpnet-base-v2 accepts text only. Both return embeddings as their only output modality, and neither lists supported parameters, so you cannot tune sampling style options on either side. The qualitative comparison marks Nemotron 3.5 Embed 8B as the one with the bigger context, and it is the only one of the two that handles images at all.

For practical work, the choice usually follows from your input. If you are embedding long documents, whole pages, transcripts or anything that comfortably exceeds a few hundred tokens, Nemotron 3.5 Embed 8B lets you send that content in one piece instead of splitting it into chunks and reconciling the results. If you need to embed images alongside text, or build a mixed index, it is the only option here that accepts image input. all-mpnet-base-v2 is a reasonable fit when your inputs are short, such as sentences, titles, short queries or individual passages, and you want a straightforward text embedding model. It is also the older of the two by provider release date, though that alone does not decide anything.

Calling either model through api14 works the same way because the API is OpenAI-compatible. You point your client at the embeddings endpoint, send your input, and set the model field to the id you want. To switch between them, change only that model id, from nvidia/nemotron-3.5-embed-8b:free to sentence-transformers/all-mpnet-base-v2 or the reverse. The request shape, authentication and response handling stay the same, so you can move an existing pipeline over by editing one string. The live price table on this page shows what each call costs, so check it there rather than assuming.

There are still cases where all-mpnet-base-v2 is the better pick. If every input you embed is short text well under 512 tokens, its smaller context is not a limitation, and you avoid paying for capacity you will not use. It is also the simpler choice when you specifically want a text-only embedding model and have no image content in your pipeline. Nemotron 3.5 Embed 8B remains the pick when context length or image input matters, but for short, plain text workloads the smaller model is entirely adequate.

Switch with one string

Both models use the same endpoint and request shape. Change the model value from "nvidia/nemotron-3.5-embed-8b:free" to "sentence-transformers/all-mpnet-base-v2" and prompts, streaming and tools keep working.

NVIDIA: Nemotron 3.5 Embed 8B (free) details & playground →Sentence Transformers: all-mpnet-base-v2 details & playground →

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api14.com/api/v1",
  apiKey: process.env.API14_API_KEY,
});

const result = await client.embeddings.create({
  "model": "sentence-transformers/all-mpnet-base-v2",
  "input": [
    "The quick brown fox",
    "Lorem ipsum"
  ]
});
console.log(result.data.map((d) => d.embedding.length));

More comparisons

Nemotron 3.5 Embed 8B (free) vs all-mpnet-base-v2: API price, context & features · api14