voyage-3.5-lite
Voyage 3.5 Lite updates the smaller third-generation encoder. Its training changes include distillation from larger retrieval models.
Model design
Voyage reports unchanged model size relative to voyage-3-lite. The update supports Matryoshka dimensions and quantization-aware output with a 32,000-token input limit.voyage-3.5 and voyage-3.5-lite
Practical use
Practical guidance. A smaller encoder is useful when query processing is frequent. Check long and multilingual inputs independently before replacing a working model. Select vector dimensions using retrieval measurements and the available index memory.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Launch date | |
| Access | API |
| Atlas lifecycle | Legacy since 15 January 2026MongoDB Voyage AI model lifecycle: Atlas scope and release history |
| Output dimensions | 1,024 default; 256, 512 or 2,048 optionalMongoDB Voyage AI model specifications |
Related models
- voyage-4-lite
Voyage AI · Voyage 4
Smaller hosted text encoder in the shared Voyage 4 family.
Launch date API - voyage-3.5
Voyage AI · Voyage 3.5
Updated general embeddings distilled from larger retrieval models.
Launch date API - voyage-3-lite
Voyage AI · Voyage 3
Smaller third-generation encoder with 512-dimensional output.
Launch date API