text-embedding-3-small vs Voyage 4 Lite
Which one to choose
Pick OpenAI's text-embedding-3-small for a compact, economical general-purpose embedding model covering search, clustering, recommendations and classification, as an upgrade on ada. Pick Voyage 4 Lite when query volume and latency dominate and your documents are indexed with Voyage 4 or Voyage 4 Large, since its vectors are compatible with both. Voyage 4 Lite suits retrieval pipelines; text-embedding-3-small suits all-purpose indexes.
Pricing comparison
| text-embedding-3-small | Voyage 4 Lite | |
|---|---|---|
| Model maker | OpenAI | Other |
| Delivery availability | Available | Available |
| Context window | 8.2K | 32K |
| Max output | - | - |
| Official price | Input$0.02per 1M tokensOutput$0.00per 1M tokens | Input$0.02per 1M tokensOutput$0.00per 1M tokens |
| TokenLab price | Input$0.014per 1M tokensOutput$0.00per 1M tokens | — |
| Model performance | Collecting data | Collecting data |
| Capabilities |
Choose text-embedding-3-small when
- You run a first search index or RAG prototype and want a simple default
- You also classify, cluster or recommend with the same vectors
- You upgrade an index built on ada embeddings
Choose Voyage 4 Lite when
- You embed many live queries and care about response time
- Your document index comes from Voyage 4 or Voyage 4 Large
- You want a selectable vector size from the same family
How they differ
| Aspect | text-embedding-3-small | Voyage 4 Lite |
|---|---|---|
| Role | OpenAI's compact model, positioned as the economical option beside the large one. | Voyage's latency- and cost-oriented model within its fourth series. |
| Family compatibility | Its vectors stand apart from OpenAI's large model. | Its vectors match Voyage 4 and Voyage 4 Large, so queries can use it against a larger model's index. |
| Task range | Search, clustering, recommendations, anomaly detection and classification. | Retrieval-focused, not specialised for code, finance or legal text. |
| Quality tier | Below text-embedding-3-large on precise matching. | Below Voyage 4 and Voyage 4 Large in retrieval quality. |
| Vector size options | Dimensions can be shortened. | Four sizes are available, small to large. |
Summary
- text-embedding-3-small: Input $0.014 / Output $0.00 per 1M tokens; Voyage 4 Lite: Input $0.02 / Output $0.00 per 1M tokens. 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.
- Voyage 4 Lite has a 3.9x larger context window
FAQ
Which is better for a RAG chatbot, text-embedding-3-small or Voyage 4 Lite?
If you build the whole index with one model, either works; start with text-embedding-3-small for simplicity. If you want higher-quality document vectors and fast queries, index with Voyage 4 Large and query with Voyage 4 Lite.
Can Voyage 4 Lite read an index built with text-embedding-3-small?
No. Vectors from different providers are not comparable, so the existing index must be rebuilt with Voyage 4 Lite before queries can be matched against it.
Which one is the newer model?
Voyage 4 Lite is from Voyage's fourth series and succeeds Voyage 3.5 Lite. text-embedding-3-small is OpenAI's improvement over the ada model; the two have no upgrade relationship.
Which is cheaper, text-embedding-3-small or Voyage 4 Lite?
text-embedding-3-small: Input $0.014 / Output $0.00 per 1M tokens; Voyage 4 Lite: Input $0.02 / Output $0.00 per 1M tokens. 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.
What are the key differences between text-embedding-3-small and Voyage 4 Lite?
Both models share similar capabilities.
Sources
Reviewed Oct 2, 2026