Skip to content

Voyage 4 Lite

Voyage 4 Lite is the lightweight Voyage 4 model. Voyage AI reports it approaches voyage-3.5 retrieval accuracy with fewer parameters, shares one embedding space with voyage-4-large and voyage-4, and supports a context window of 32K tokens with Matryoshka dimensions and quantization like the rest of the family.

index.ts
import { embed } from 'ai';
const result = await embed({
model: 'voyage/voyage-4-lite',
value: 'Sunny day at the beach',
})

Frequently Asked Questions

  • What is the difference between Voyage 4 Lite and voyage-4?

    voyage-4 is the mid-sized Voyage 4 model; Voyage AI reports it approaches voyage-3-large quality. Voyage 4 Lite uses fewer parameters; Voyage AI reports it approaches voyage-3.5 accuracy. Both share the same embedding space as voyage-4-large.

  • How does Voyage 4 Lite compare to voyage-3.5-lite?

    Voyage 4 Lite is a Voyage 4 model with a shared embedding space and updated training. Voyage AI positions it near voyage-3.5 accuracy with fewer parameters. Moving from Voyage 3.x requires re-embedding.

  • What is the context window for Voyage 4 Lite?

    32K tokens. Size chunks so single requests stay under this limit.

  • Is Voyage 4 Lite suitable for production use?

    Yes. Voyage AI targets production for voyage-4-lite, including asymmetric setups with voyage-4-large document embeddings when query cost matters.

  • How do I access Voyage 4 Lite through Vercel AI Gateway?

    Add your Voyage AI API key in AI Gateway settings, then send embedding requests through AI Gateway. AI Gateway authenticates requests and records usage.

  • When should I move from Voyage 4 Lite to voyage-4 or voyage-4-large for queries?

    Move when you need higher query-side accuracy. If documents stay on voyage-4-large, you can upgrade query embeddings to voyage-4 or voyage-4-large without re-vectorizing documents in that asymmetric setup.

  • Do I need to re-embed my data to switch from voyage-3.5-lite?

    Yes. Voyage 3 and Voyage 4 use different embedding spaces, so you re-embed when you change generations.