A solid choice for search and RAG builders who need text, images and video in one vector space, with a free tier to test on.
Kenning's take
What it does
Gemini Embedding 2 turns text, images, video, audio and documents into vectors in a single shared space, so search and retrieval can work across media types. It handles over 100 languages and lets developers shrink vector size to save storage. It launched in public preview in March 2026 and became a stable model in April.
- Text, image, video, audio and PDF input
- Mixed inputs in one request
- 100+ languages
- Flexible output dimensions up to 3072
- Batch pricing at half rate
Who it suits. Developers building search, RAG and recommendation systems.
Key facts
- Category
- AI models
- Part of
- Google Gemini
- Made by
- Based in
- United States
- Released
- 10 March 2026
- How to get it
- Gemini API and Vertex AI (now Gemini Enterprise Agent Platform)
- Works on
- API
- In the UK
- Yes
- Free plan
- Yes, Gemini API free tier
- API price
- $0.20 per million text tokens, $0.45 image, $6.50 audio, $12.00 video (batch half price)
FAQ
When was Gemini Embedding 2 released?
In public preview on 10 March 2026; Google's deprecations table lists the stable model from 22 April 2026.
What can it embed?
Text, images, video, audio and PDFs, including mixed inputs.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
Gemini 2.5 Flash Image (Nano Banana)
Google's 2025 image generation and editing model, known as Nano Banana, now being retired
Gemini 3
Google's third generation of Gemini models, launched November 2025 with Gemini 3 Pro
Gemini 3 Flash
Faster, cheaper Gemini 3 model released December 2025 for everyday tasks and agents
Gemini 3.1 Deep Think
Gemini's slowest, most thorough reasoning mode for science, maths and engineering problems