Embedding models for search and RAG
RAG and semantic search start with a solid embedding. On Geek Hub you get text-embedding-3-small (cheap and strong), 3-large (higher quality), and ada-002 (legacy) with the same billing as the rest of the catalog.
Models in this collection
| Model | Type | Context | Price |
|---|---|---|---|
 OpenAI text-embedding-3-smallOpenAI | Embedding | 8k tokens | $0.0218 / 1M tokens |
| OpenAI text-embedding-3-largeOpenAI | Embedding | 8k tokens | $0.1417 / 1M tokens |
| OpenAI text-embedding-ada-002OpenAI | Embedding | 8k tokens | $0.109 / 1M tokens |
Why these models
3-small is the cost default; 3-large when retrieval quality matters more than pennies; ada-002 only if a legacy index requires it.
Use them with Geek Hub
One OpenAI-compatible base URL and API key. Swap any model id from this list without rewriting your client.
Get an API keyFAQ
- Which embedding for RAG?
- Start with text-embedding-3-small. Move to 3-large if recall is not enough. Do not mix dimensions in the same index.
- Same API key as chat?
- Yes. Switch the model id to the embedding you need.
- How is it billed?
- Per 1M input tokens. The table shows the representative price.