RAG → embedding APIs
What do embeddings cost per 1M tokens?
82 embedding endpoints from 16 providers, from $0.004 to $0.18 per 1M tokens: a 45× spread for the same job. Enter your monthly volume below.
- Cheapest$0.004 / 1M tokensPplx embed V1 0.6b · Perplexity
- Best retrieval score (67.3)$0.01 / 1M tokensQwen3 embedding 8b · Nebius Token Factory
- Most expensive$0.18 / 1M tokensVoyage context 3 · Voyage AI
Standard list prices from each provider's pricing page, checked daily. Last price change Oct 19, 2026. How we compare.
Ranked for your volume
Monthly cost by provider
| Endpoint | Estimated / month | Retrieval score | Per 1M tokens |
|---|---|---|---|
1 PerplexityPplx embed V1 0.6b - lowest for your volumePricing page | $0.40 | Not benchmarked | $0.004 |
2 Google Cloud Vertex AIText embedding preview 0409Pricing page | $0.63 | Not benchmarked | $0.0063 |
3 Fireworks AIGte basePricing page | $0.80 | Not benchmarked | $0.008 |
4 Fireworks AINomic embed text V1Pricing page | $0.80 | 46.5 | $0.008 |
5 Fireworks AINomic embed text v1.5Pricing page | $0.80 | Not benchmarked | $0.008 |
6 PerplexityPplx embed context V1 0.6bPricing page | $0.80 | Not benchmarked | $0.008 |
7 Together AIBge base en v1.5Pricing page | $0.80 | Not benchmarked | $0.008 |
8 Together AITogether AI embedding up to 150mPricing page | $0.80 | Not benchmarked | $0.008 |
9 Nebius Token FactoryBge en iclPricing page | $1.00 | 61.8 | $0.01 |
10 Nebius Token FactoryBge multilingual gemma2Pricing page | $1.00 | 57.2 | $0.01 |
Ordered by listed price for your volume only. No provider pays for its position. Share this page URL to keep your numbers.
Retrieval score: mean nDCG@10 on ArguAna, FiQA2018, SCIDOCS, TRECCOVID from the public MTEB results (CC0), commit 964b762 (source), checked Oct 31, 2026. Higher is better; unscored models have no published result for the exact model.
Before you pick
Text-embedding input prices; multimodal embeddings are not included.
Retrieval score: our average of nDCG@10 on four English MTEB retrieval tasks (ArguAna, FiQA, SCIDOCS, TREC-COVID), from the public MTEB results (CC0), for the exact model; higher is better. Results differ by language and domain, so test on your own documents. Smaller dimensions can also cut vector-storage cost.
Batch discounts are not applied.
At the default volume (100 M tokens a month) the cheapest endpoint costs $0.40 and the most expensive $18.00.
Keep comparing
Related AI API prices
Web search API prices for agents - LLM API prices - Monthly price reports