# Fireworks releases Voyage Multimodal 3.5 embedding model

> The new model can embed text, images, tables, slides and video frames for retrieval, with up to 32K token context and flexible embedding sizes.

Oossa · 2026-10-06 · https://oossa.com/en/fireworks-releases-voyage-multimodal-3-5-embedding-model

Fireworks AI announced Voyage Multimodal 3.5, a retrieval‑focused embedding model that handles interleaved text and visual content. It works with screenshots, PDFs, tables, figures, slides and video frames. The model supports up to 32 000 tokens, counting each 1120 pixels of video as one token. Users can choose embedding dimensions of 256 to 2048, thanks to Matryoshka embeddings, and select from several quantization formats to cut storage costs.

## The facts

- Supports up to 32 000 tokens (video frames count as 1 token per 1120 pixels).
- Embeddings available in 256, 512, 1024, or 2048 dimensions.

## Why it matters

Teams that need to search mixed‑media datasets can use Voyage Multimodal 3.5 to find relevant text or video clips without building separate indexes.

## Sources & references

1. [Voyage Multimodal 3.5](https://fireworks.ai/models/fireworks/voyage-multimodal-3-5) – Fireworks AI

Last updated: 2026-10-06