Perplexity: pplx-embed-v1-4B

Perplexity Embed v1 4B maps independent text and queries into vectors for semantic retrieval and RAG. It supports a 32K context window and returns 2560-dimensional vectors through the Embeddings API; its low input-token price suits large document indexes.

Compare models

pplx-embed-v1-4b

AvailablePerplexityEmbeddingSync

Input / Output: $0.03 / $0.00
Context: 32K
Modalities: Embedding

About pplx-embed-v1-4B

pplx-embed-v1-4B is the larger of Perplexity's two pplx-embed-v1 text embedding models, aimed at web-scale retrieval where ranking quality matters most. It maps a passage or query to a 2560-dimensional vector, accepts both without instruction prefixes, and ships with open MIT-licensed weights. Against the 0.6B sibling it spends more compute and storage per text in exchange for more capacity.

Where it works well

Outputs 2560-dimensional vectors, giving each text more room to separate fine differences than the 1024-wide 0.6B model.
Plain-text input on both sides: the model card states no instruction prefix is needed for documents or queries.
Open weights under the MIT license on a Qwen3 base, so results can be reproduced and audited.
Offers INT8 and binary output options, which can cut index size when the full-width float vectors are too large.

When to choose another model

Wide vectors make indexes several times larger than 0.6B vectors for the same corpus, which matters at hundreds of millions of documents.
Each text is embedded alone, so context from surrounding chunks of the same document is not captured.

Getting started

Create API key
Create a key in Console, then use it with every model on the platform.

Send your first request

Copy the example for your language and run it against the endpoint.

EmbeddingTokenLab endpoint

POST/v1/embeddings

curl -X POST "https://api.tokenlab.sh/v1/embeddings" \
  -H "Authorization: Bearer sk-xxx" \
  -H "Content-Type: application/json" \
  -d '{
  "encoding_format": "float",
  "model": "pplx-embed-v1-4b",
  "input": "The quick brown fox jumps over the lazy dog."
}'

Parameters

Embedding

Parameter	Values and limits	Default	Notes
`encoding_format`	Not provided	Not provided	This field is accepted. The current public contract does not specify a complete value range or list. No additional values are promised. The omitted-field default is not verified.
`input`	Not provided	Not provided	Not provided

Pricing

The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.

Embedding

per 1M tokens

Official price: Input $0.03 / Output $0.00
Verified price: —
Discount: —

	Official priceper 1M tokens	Verified priceper 1M tokens	Discount
Embedding	Input $0.03 / Output $0.00	—	—

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours

Requests
Success rate
P95 latency
Total tokens

Model performance

Last updated Nov 15, 2026, 1:51 AM UTC

Metrics appear once privacy and data volume thresholds are met.

Data is based on aggregate user requests, excluding status checks.

Open in Console

Open pplx-embed-v1-4B in Console with a prompt ready to edit or send.

Help me try pplx-embed-v1-4b with embedding at /v1/embeddings. Show the result and cost.

Use cases

High-precision search
Power the main semantic search of a knowledge base or documentation site where ranking quality outweighs index size.
RAG over mixed content
Retrieve passages for answer generation across multilingual or varied web content, with fewer misses than a lighter embedding.
Deduplication and clustering
Group near-duplicate pages or support tickets by cosine similarity of their wider vectors.

Prompt examples

Which clause lets either party terminate the contract for convenience?

kubernetes pod stuck in CrashLoopBackOff after config map change

Comment configurer l'authentification à deux facteurs sur le tableau de bord

This model has conditional pricing. Monthly token totals alone cannot produce a reliable estimate; use the detailed pricing for the request specification, cache, and applicable time window.

FAQ

What exactly is pplx-embed-v1-4B?

It is the 4-billion-parameter text embedding model in Perplexity's pplx-embed-v1 family. It converts text into dense vectors for semantic search, retrieval-augmented generation, clustering, and similarity matching, and its weights are published under the MIT license.

What is the vector size?

The model card gives 2560 dimensions for the 4B model, compared with 1024 for the 0.6B model. Plan vector storage and index memory for the wider size before you commit to a corpus-wide re-embedding.

Does it need instruction prefixes?

No. You send documents and queries as plain text. Perplexity states it avoids instruction prefixes on purpose, so a single code path can embed both sides of the search.

Can I reduce storage?

The card lists INT8 and binary quantized outputs. They shrink each stored vector substantially, so you can trade a small amount of accuracy for a much smaller index. Test recall on your own queries before switching.

4B or 0.6B for my project?

Choose 4B when answer quality depends on fine-grained ranking and the corpus fits comfortably in memory. Choose 0.6B when you embed enormous volumes or need the cheapest per-text compute.

How much does pplx-embed-v1-4B cost?

On TokenLab, pplx-embed-v1-4B costs Input $0.03 / Output $0.00 per 1M tokens. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.

Which endpoint should pplx-embed-v1-4B use?

Use https://api.tokenlab.sh/v1/embeddings for pplx-embed-v1-4B. The request example below shows the matching code shape.

Which operations does pplx-embed-v1-4B support?

pplx-embed-v1-4B supports Embedding. Select an operation above to see its endpoint and request example.