跳转到主要内容
GoogleApr 17, 2026

Gemini Embedding 2 Preview

gemini-embedding-2-previewAnonymized

Gemini Embedding 2 Preview is a preview of Google’s Gemini-based embedding model, returning 3,072-dimensional vectors for inputs of up to 2,048 tokens. It’s built for semantic search, retrieval and classification.

$0.25
per 1M input tokens
3,072
dimensions, or fewer
2,048
tokens per input
How it comparesamong 9 embedding models
Pricecheaper than 0%$0.25
Max inputlonger than 13%2,048 tokens

Further right is better. Hover a dot for its model.

Quickstart

OpenAI-compatible: point any OpenAI SDK at the Venice base URL with a Venice API key.

curl https://api.venice.ai/api/v1/embeddings \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-embedding-2-preview",
"input": "Private AI for everyone."
}'
Base URL https://api.venice.ai/api/v1

Parameters

ParameterValuesNotes
modelgemini-embedding-2-previewRequired
inputA string, or up to 2,048 stringsRequired, up to 2,048 tokens each
dimensionsUp to 3,072Returns a shorter vector
encoding_formatfloat | base64Default float; base64 is a smaller payload

LangChain’s OpenAIEmbeddings sends token IDs, which Venice rejects. Set check_embedding_ctx_length=False.

Vectors

Each input comes back as 3,072 numbers, or fewer when you ask for them with dimensions. Stored as float32, each number takes 4 bytes.

Each square is 64 dimensions, and each row 1,024.

Dimensions
3,072dimensions per vector
12.3 GBfor 1,000,000 vectors as float32

Similarities from different models aren’t comparable, so store the model ID and dimensions with each vector.

Pricing

Billed per 1M input tokens, in USD.

Input$0.25per 1M tokens

Estimate a corpus

50Mtokens
$12.50to embed them once

Specs

Model ID
gemini-embedding-2-preview
Endpoint
POST /api/v1/embeddings
Dimensions
3,072, or fewer with dimensions
Max input
2,048 tokens per input
Batch
Up to 2,048 inputs per request
Encodings
float or base64
Input
Text only
Privacy
Anonymized
Weights
Closed
Added
Apr 17, 2026

Compare

Every embedding model on Venice, cheapest first.

ModelDimensionsMax inputPrivacyPer 1M tokens
BGE-EN-ICL4,0968,192Private$0.0125
Multilingual E5 Large Instruct1,024512Private$0.0125
Nemotron Embed VL 1B v22,0488,192Private$0.0125
Qwen3 Embedding 0.6B1,024or fewer8,192Private$0.0125
Qwen3 Embedding 8B4,096or fewer8,192Private$0.0125
Text Embedding 3 Small1,536or fewer8,191Anonymized$0.025
BGE-M31,0248,192Private$0.15
Text Embedding 3 Large3,072or fewer8,191Anonymized$0.1625
Gemini Embedding 2 Preview3,072or fewer2,048Anonymized$0.25