pplx Embed v1 4B

Feb 2026
pplx Embed v1 4B
Compare in
on
$0.03per 1M tokens
69.7MTEB multilingual v2
Feb 2026released
2,560Dimensions
33K tokensMax input

Ranking?

Full ranking →
1F2LLM v2 14B80.82C2LLM 7B80.73Seed1.6 Embedding 121580.74Qwen 3 Embedding 8B80.75F2LLM v2 8B80.2
···
9F2LLM v2 1.7B78.810pplx Embed v1 4B78.711Giga Embeddings Instruct 10B A1.8B 082678.4

Run it yourself

estimate
Fits on✓ 8 GB (phone, laptop) · 8-bit✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceMITfree for commercial use

Benchmarks

in Embeddings & Retrieval
CoIR · Mean over tasks · multilingual (9 languages)77.3#14
CoIR · Mean over task types · multilingual (9 languages)77.3#14
CoIR · Retrieval · multilingual (9 languages)77.3#14
MTEB Code v1 · Mean over tasks · multilingual (15 languages)78.7#10
MTEB Code v1 · Mean over task types · multilingual (15 languages)78.7#10
MTEB Code v1 · Retrieval · multilingual (15 languages)78.7#10
MTEB multilingual v2 · Retrieval · multilingual (120 languages)69.7#8

FAQ

How much does pplx Embed v1 4B cost?›

$0.03 per 1M tokens.

Is pplx Embed v1 4B free to use?›

There is no free API; it costs $0.03 per 1M tokens. The weights are open, so you can run it yourself.

How good is pplx Embed v1 4B, and what is it best at?›

#8 of 174 in Embeddings.

Is pplx Embed v1 4B better than F2LLM v2 1.7B?›

On MTEB Code v1 · Mean over tasks · multilingual (15 languages), pplx Embed v1 4B scores 78.7 vs 78.8 — about the same.

When was pplx Embed v1 4B released? Is there a newer version?›

Released 2026-02-11. It is the current version in its line.

Is pplx Embed v1 4B open source?›

Yes, the weights are open (MIT licence).

Can I run pplx Embed v1 4B locally?›

The weights are open and can be self-hosted; hardware needs are not confirmed yet.

What is pplx Embed v1 4B?›

pplx Embed v1 4B is an open-weights embeddings & retrieval model from perplexity-ai, released Feb 2026.

What are the best alternatives to pplx Embed v1 4B?›

Qwen 3 Embedding 8B (cheaper), F2LLM v2 14B (better), C2LLM 7B (open weights).