Voyage AI Pricing in 2026: Embeddings, Rerankers and the 200M Free Tokens
What every current Voyage AI embedding and rerank model costs per 1M tokens, how the free token allowance works, and how Voyage compares with OpenAI, Google and Cohere embeddings.
Voyage AI charges per token: $0.02 per 1M tokens for voyage-4-lite, $0.06 for voyage-4 and $0.12 for voyage-4-large, with rerankers at $0.02 (rerank-3-lite) and $0.05 (rerank-3). Every account gets the first 200 million tokens free on each current model. The cheapest option is voyage-4-lite, which matches OpenAI’s text-embedding-3-small at $0.02 and costs a sixth of voyage-4-large.
All prices below come from Voyage AI’s official pricing page and our Voyage AI provider page, as of November 2026. Check the live page before committing to a large indexing job.
How much does Voyage AI cost?
Voyage bills only input tokens. Embeddings and reranking have no output token charge, and there is no subscription or monthly fee. You pay for what you embed or rerank, at the rate for the model you call.
- Per-token billing for text embeddings and rerankers, priced per 1M tokens.
- Pixels plus tokens for multimodal embeddings.
- 33% off with the Batch API, which returns results within 12 hours.
- Free tokens per account on current models (see below). Older models get no free tokens.
Voyage AI embedding model prices
| Model | Price per 1M tokens | Free tokens | Best for |
|---|---|---|---|
| voyage-4-large | $0.12 | 200M | Highest retrieval quality, multilingual |
| voyage-4 | $0.06 | 200M | General-purpose retrieval |
| voyage-4-lite | $0.02 | 200M | Lowest cost and latency |
| voyage-context-4 | $0.12 | 200M | Contextualized chunk embeddings |
| voyage-code-4 | $0.12 | 200M | Code retrieval and coding agents |
| voyage-finance-2, voyage-law-2, voyage-code-2 | $0.12 | 50M | Domain-specific retrieval |
The voyage-4 models have a 32,000-token context window and output 1024 dimensions by default, with 256, 512 and 2048 also available. Voyage says all embeddings from the 4 series are compatible with each other. That means you can embed your documents once with voyage-4-large and embed queries with the cheaper voyage-4-lite without re-indexing.
Multimodal embeddings (voyage-multimodal-3.5 and voyage-multimodal-3) cost $0.12 per 1M text tokens plus $0.60 per billion pixels. Each image is billed as at least 50,000 and at most 2 million pixels, so one image costs between $0.00003 and $0.0012. Each video frame counts as an image.
How much does voyage-3.5 cost?
voyage-3.5 still costs $0.06 per 1M tokens, the same as voyage-4, and voyage-3.5-lite costs $0.02. Voyage now lists both as older models, and older models get no free tokens. voyage-3-large, voyage-code-3 and voyage-context-3 cost $0.18 per 1M, which is more than voyage-4-large. If you are starting a new index, the voyage-4 equivalents cost the same or less and come with the free allowance.
Is Voyage AI free?
Partly. The pricing page says the first 200 million tokens for voyage-4-large, voyage-4, voyage-4-lite and voyage-context-4 are free for every account. voyage-finance-2, voyage-law-2 and voyage-code-2 get 50 million. Multimodal models get 200M text tokens and 150 billion pixels free. Current rerankers get 200M free tokens.
Some details are worth knowing:
- It’s per account, not monthly. Voyage describes it as “the first 200 million tokens”. The page doesn’t mention a monthly reset.
- It’s per model. Each model has its own allowance, so you can test several models before paying.
- Batch usage doesn’t use it. Free tokens don’t apply to Batch API requests, and batch tokens don’t come out of your free balance.
- Older models are excluded, including voyage-3.5 and voyage-3-large.
For scale, 200 million tokens is roughly 150,000 documents of 1,000 words each. Most prototypes and small RAG apps never leave the free tier. See our AI API free tier guide for how this compares with other providers.
How much does Voyage rerank cost?
| Model | Price per 1M tokens | Est. cost per request* | Free tokens |
|---|---|---|---|
| rerank-3 | $0.05 | $0.0025 | 200M |
| rerank-3-lite | $0.02 | $0.001 | 200M |
*Voyage’s estimate assumes 100 documents per request and 500 tokens for the query plus each document.
Billed tokens are (query tokens × number of documents) + the total document tokens, so a long query costs more when you rerank many documents. Both rerankers have a 32,000-token context. The older rerank-2.5 and rerank-2 cost the same ($0.05 and $0.02 for the lite versions), but Voyage says the rerank-3 models beat them on quality, context, latency and throughput.
Voyage AI vs OpenAI, Google and Cohere embedding pricing
| Model | Provider | Price per 1M tokens | Free allowance |
|---|---|---|---|
| BGE-M3 (open-weight) | DeepInfra | $0.01 | — |
| voyage-4-lite | Voyage AI | $0.02 | 200M tokens |
| text-embedding-3-small | OpenAI | $0.02 | — |
| voyage-4 | Voyage AI | $0.06 | 200M tokens |
| voyage-4-large | Voyage AI | $0.12 | 200M tokens |
| text-embedding-3-large | OpenAI | $0.13 | — |
| Gemini Embedding 2 | $0.20 ($0.10 batch) | Gemini API free tier |
A few points stand out:
- voyage-4-large is cheaper than OpenAI’s large model: $0.12 vs $0.13 per 1M, plus 200M free tokens.
- Gemini Embedding 2 costs the most for text at $0.20, but it’s free on the Gemini API free tier (with rate limits), and it embeds images, audio and video too.
- Open-weight models are cheapest. BGE-M3 costs $0.01 per 1M on DeepInfra. That’s half the price of voyage-4-lite before free tokens, though you lose Voyage’s retrieval-tuned quality.
- Cohere Embed 4: Cohere’s pricing page no longer shows a self-serve per-token rate. Its free trial key costs nothing but is rate-limited and not allowed for production use.
The cheapest embedding APIs ranking lists every provider we track.
Example: embedding 1 billion tokens
| Model | Cost |
|---|---|
| voyage-4-lite | $16 (800M paid after 200M free) |
| voyage-4-large | $96 (800M paid after 200M free) |
| text-embedding-3-large | $130 |
| Gemini Embedding 2 (standard) | $200 |
Does the Voyage Batch API save money?
Yes. The Batch API is 33% cheaper than the standard endpoints, with a 12-hour completion window. That brings voyage-4-large to about $0.08 per 1M tokens, voyage-4 to about $0.04 and voyage-4-lite to about $0.013. Batch tokens can’t use the free allowance, so use the standard endpoint until the free tokens run out, then switch bulk indexing to batch. File storage for the Files API costs $0.05 per GB per month, and files are deleted after 30 days.
Fine-tuned Voyage models on dedicated instances are priced through sales.
Which Voyage model should you use?
- Prototyping or small apps: any voyage-4 model. The free 200M tokens will likely cover you.
- Cost-sensitive production at scale: voyage-4-lite at $0.02, or index with voyage-4-large and query with voyage-4-lite, since the 4 series share one embedding space.
- Best retrieval quality: voyage-4-large plus rerank-3. Reranking the top 100 results costs about $0.0025 per query.
- Code search: voyage-code-4 at $0.12.
For a full RAG cost estimate, add your LLM’s cost: see the cheapest LLM APIs ranking.