Voyage AI Models
vai supports the full range of Voyage AI embedding, reranking, and multimodal models. This page provides a complete reference.
Current Embedding Models
| Model | Architecture | Context | Dimensions | Price/1M tokens | RTEB Score |
|---|---|---|---|---|---|
voyage-4-large | MoE | 32K | 256, 512, 1024, 2048 | $0.12 | 71.41 |
voyage-4 | Dense | 32K | 256, 512, 1024, 2048 | $0.06 | 70.07 |
voyage-4-lite | Dense | 32K | 256, 512, 1024, 2048 | $0.02 | 68.10 |
Bold = default dimensions. All Voyage 4 models support flexible dimensions via Matryoshka representation learning.
Shared Embedding Space
All Voyage 4 models share the same embedding space, enabling asymmetric retrieval — embed documents cheaply, queries with the best model.
Domain-Specific Models
| Model | Context | Dimensions | Price/1M tokens | Domain |
|---|---|---|---|---|
voyage-code-3 | 32K | 256, 512, 1024, 2048 | $0.18 | Code retrieval |
voyage-finance-2 | 32K | 1024 | $0.12 | Financial documents |
voyage-law-2 | 16K | 1024 | $0.12 | Legal documents |
Multimodal Model
| Model | Context | Dimensions | Price | Modalities |
|---|---|---|---|---|
voyage-multimodal-3.5 | 32K | 256, 512, 1024, 2048 | $0.12/M tokens + $0.60/B pixels | Text, images, video |
Reranking Models
| Model | Context | Price/1M tokens | Best For |
|---|---|---|---|
rerank-2.5 | 32K | $0.05 | Best quality reranking |
rerank-2.5-lite | 32K | $0.02 | Fast, cost-effective |
Upcoming Models
| Model | Type | Notes |
|---|---|---|
voyage-4-nano | Embedding | Open-weight, free, runs locally. 512d default. |
voyage-context-3 | Embedding | Contextualized chunk embeddings |
Legacy Models
Use vai models --all to see legacy models. These include voyage-3-large, voyage-3.5, voyage-3.5-lite, voyage-code-2, voyage-multimodal-3, rerank-2, and rerank-2-lite.
Competitive Landscape (RTEB Benchmarks)
| Model | Provider | NDCG@10 Score |
|---|---|---|
| voyage-4-large | Voyage AI | 71.41 |
| voyage-4 | Voyage AI | 70.07 |
| Gemini Embedding 001 | 68.66 | |
| voyage-4-lite | Voyage AI | 68.10 |
| Cohere Embed v4 | Cohere | 65.75 |
| OpenAI v3 Large | OpenAI | 62.57 |
Browse Models in vai
# List all current models
vai models
# Show benchmarks
vai models --benchmarks
# Include legacy models
vai models --all
# JSON output
vai models --json
Further Reading
- Choosing a Model — Decision guide
- Voyage 4 Family — Deep dive into the Voyage 4 lineup
- Domain-Specific Models — When to use specialized models