Generate an embeddings request plus a cosine-similarity helper.
—
Plugsky is OpenAI-compatible — see docs and start on the free plan (2 free models, no card).
The Embedding API Tester compares embedding endpoints from different providers on the things that matter for retrieval: vector dimensions, response latency, and quality on your own text. You send sample content and inspect the vectors returned. It is aimed at developers building search, recommendation, or RAG pipelines who need to choose an embedding model before indexing a corpus. Plugsky serves embedding models through an OpenAI-compatible endpoint.
Dimensions, latency, price per token, maximum input length, and retrieval quality on your domain. Quality matters most: run a small labeled set of queries and documents, and measure whether the right passage ranks near the top.
Storage and memory grow with dimensions, as do vector database costs and search time. Higher dimensions do not automatically mean better retrieval, so test a smaller model first and only move up if quality measurably improves.
Not without re-embedding. Vectors from different models are not comparable, and changing dimensions breaks the index. Fix the model and version before you index at scale, and keep the source text so you can re-embed later.
Canonical pricing and plans: plugsky.com/#sec-pricing · Terms · SLA · Docs