Gemini embedding-001 pricing at a glance
A current, route-specific Google price for gemini-embedding-001 is not confirmed by the sources cited here. Google’s July 14, 2025 announcement stated $0.15 per 1 million input tokens, without separate Gemini API and Vertex AI rates. An independent pricing page reports Vertex AI rates of $0.15 per million tokens online and $0.12 per million for batch, checked October 16, 2026; these are third-party-reported figures, not official confirmation. Google’s current Gemini API documentation says gemini-embedding-001 remains available for text-only use, but the current price for that route is not established here.
| Route or mode | Rate | Evidence and scope |
|---|---|---|
| Gemini API | $0.15 / 1 million input tokens | Historical rate in Google’s July 14, 2025 announcement; not confirmed as the current rate or as a separate route-specific tariff. |
| Vertex AI, online | $0.15 / 1 million tokens | Reported by an independent pricing page checked October 16, 2026; not confirmed here by a current official price listing. |
| Vertex AI, batch | $0.12 / 1 million tokens | Reported by the same independent page; current official price and terms are not confirmed here. |
The model-specific pricing excerpts do not state a currency code. Do not treat the reported Vertex AI batch rate as a Gemini API Batch API rate, or infer a model-specific free allowance or output-token charge from general pricing terms.
Estimate your embedding bill
Use the input-token volume and the rate applicable to your route and processing mode:
Estimated charge = input tokens ÷ 1,000,000 × rate per 1,000,000 tokens
Using the reported Vertex AI rates as planning examples—not official quotes:
| Input-token volume | Online at reported $0.15/M | Batch at reported $0.12/M |
|---|---|---|
| 1 million | $0.15 | $0.12 |
| 10 million | $1.50 | $1.20 |
| 100 million | $15.00 | $12.00 |
These are arithmetic estimates based on third-party-reported rates. They assume all listed tokens are billable input tokens at that rate; they do not establish taxes, account terms, or other charges. Confirm the applicable rate and terms for your Google billing route before using an estimate as a production commitment.
Which API route and terms should you budget for?
Google’s 2025 announcement described gemini-embedding-001 as generally available through Gemini API and Vertex AI and gave one price, without publishing separate route-specific rates. Current Google Gemini API documentation says the model remains available for text-only use. The cited sources do not confirm a current Gemini API price for the model; the Vertex AI online and batch rates above remain third-party reports.
The announcement described Gemini API free and paid tiers generally, but does not establish a current model-specific free allowance or quota for gemini-embedding-001. Likewise, do not assume batch availability or a discount on one route based on terms reported for another.
Production API fit and input limit
Google’s Vertex AI text-embeddings API reference lists gemini-embedding-001 with a 2,048-token maximum sequence length and says a request for this model can include only one input text. Check each text against the limit and split longer material if needed; the cited reference does not specify what happens when an input exceeds the limit.
Google’s July 2025 announcement showed this Gemini API Python SDK call. It is the announcement’s example, not a complete production setup:
from google import genai
client = genai.Client()
result = client.models.embed_content(
model="gemini-embedding-001",
contents="Text you want to embed"
)
print(result.embeddings)
The announcement described a default output dimension of 3,072 and recommended 3,072, 1,536, or 768 dimensions. It did not establish that choosing a smaller dimension changes the billed token rate.
Budget takeaway
Google’s July 14, 2025 announcement gives a historical rate of $0.15 per million input tokens. A third-party page checked October 16, 2026 reports Vertex AI rates of $0.15/M online and $0.12/M batch, but these are not current, route-specific Google quotes confirmed here. For Vertex AI workload fit, Google’s reference supports a 2,048-token maximum for one input text per request. Verify current route-specific pricing and terms before committing a production budget.
Sources
Gemini Embedding now generally available in the Gemini API
Google Gemini Embedding Pricing: gemini-embedding-2, gemini-embedding-001 & text-embedding-005