Perplexity launches high-performance embedding models

What's new? Perplexity launched pplx-embed-v1 and pplx-embed-context-v1 for large-scale retrieval; models use int8 and binary quantization in 0.6b and 4b sizes;