Voyage AI: 200M free embedding tokens (per account)
Source: https://docs.voyageai.com/docs/pricing
Description
Create account to comment on specific lines or Sign in
+ 1 Voyage AI — acquired by MongoDB in early 2024 and now branded Voyage AI by MongoDB — gives every new account 200 million free tokens per model on its flagship embeddings and rerankers, plus 50M free tokens on legacy specialized models and a dedicated 150 billion free pixels allocation for multimodal embeddings. No credit card is required to claim the free tokens, and signup is wired through MongoDB Atlas (free Atlas account, no Atlas cluster required).
No comments on this line yet.
+ 2
No comments on this line yet.
+
3
If you build RAG, semantic search, code search or hybrid retrieval, this is currently one of the largest free buckets on the market — by way of comparison, OpenAI's text-embedding-3-small would burn through 200M tokens at ~$4 of paid usage per model, and Voyage's quality on the MTEB / BEIR-style benchmarks is consistently competitive with Cohere and OpenAI.
No comments on this line yet.
+ 4
No comments on this line yet.
+
5
No comments on this line yet.
+ 6
No comments on this line yet.
+ 8
No comments on this line yet.
+ 9 1. Go to cloud.mongodb.com and sign up for a MongoDB Atlas account (email + password, or Google/GitHub SSO). MongoDB will auto-create an Organization and Project for you.
No comments on this line yet.
+ 10 2. Email-verify your account when prompted. No phone verification, no credit card.
No comments on this line yet.
+ 11 3. In your Atlas project, open the left sidebar and click "AI Models".
No comments on this line yet.
+
12
4. Click "Create model API key", give it a name (e.g. dev-key), and click Create.
No comments on this line yet.
+
13
5. Copy the API key immediately — Atlas only shows it once. Store it in a password manager or .env file.
No comments on this line yet.
+ 14 6. Export it in your shell so the official client picks it up automatically:
No comments on this line yet.
+
15
No comments on this line yet.
+ 16 export VOYAGE_API_KEY="pa-..."
No comments on this line yet.
+
17
No comments on this line yet.
+ 18 7. Install the SDK and run a smoke test:
No comments on this line yet.
+
19
No comments on this line yet.
+ 20 pip install voyageai
No comments on this line yet.
+ 21 python -c "import voyageai; print(voyageai.Client().embed(['hello'], model='voyage-4-lite').embeddings[0][:3])"
No comments on this line yet.
+
22
No comments on this line yet.
+ 23
No comments on this line yet.
+ 24 Important caveats:
No comments on this line yet.
+ 25 • You do not need to deploy a MongoDB Atlas cluster — Voyage runs as a standalone API. Atlas is just the auth/billing surface.
No comments on this line yet.
+
26
• The API is OpenAI-style REST at https://api.voyageai.com/v1/embeddings (and /rerank), so you can use requests/fetch directly without the SDK.
No comments on this line yet.
+ 27 • If you lose the API key, you cannot recover it — just create a new one and revoke the old.
No comments on this line yet.
+ 28
No comments on this line yet.
+
29
No comments on this line yet.
+ 30
No comments on this line yet.
+ 32
No comments on this line yet.
+
33
The 200M is per model, not pooled across all models. That means you can burn 200M on voyage-4-large and still have a separate 200M waiting on voyage-4, voyage-context-3, rerank-2.5, etc.
No comments on this line yet.
+ 34
No comments on this line yet.
+ 36
No comments on this line yet.
+ 37 ModelType
No comments on this line yet.
+
38
voyage-4-largeBest-quality general-purpose embeddings
No comments on this line yet.
+
39
voyage-4Balanced quality/perf embeddings
No comments on this line yet.
+
40
voyage-4-liteCheapest/fastest general embeddings
No comments on this line yet.
+
41
voyage-context-3Contextualized chunk embeddings (carries surrounding context)
No comments on this line yet.
+
42
voyage-code-3Code retrieval / dev-doc search
No comments on this line yet.
+
43
rerank-2.5Highest-accuracy reranker
No comments on this line yet.
+
44
rerank-2.5-liteLatency-optimized reranker
No comments on this line yet.
+
45
rerank-2, rerank-2-liteLegacy rerankers (still 200M)
No comments on this line yet.
+ 46
No comments on this line yet.
+ 48
No comments on this line yet.
+ 49 ModelType
No comments on this line yet.
+
50
voyage-multilingual-2Multilingual embeddings (legacy — voyage-4 is now multilingual)
No comments on this line yet.
+
51
voyage-finance-2Finance-domain embeddings
No comments on this line yet.
+
52
voyage-law-2Legal-domain embeddings
No comments on this line yet.
+
53
voyage-code-2Legacy code embeddings (use voyage-code-3 instead)
No comments on this line yet.
+ 54
No comments on this line yet.
+ 56
No comments on this line yet.
+ 57 ModelFree allocation
No comments on this line yet.
+
58
voyage-multimodal-3.5200M text tokens + 150B pixels (text + image + video)
No comments on this line yet.
+
59
voyage-multimodal-3200M text tokens + 150B pixels (legacy, text + image only)
No comments on this line yet.
+ 60
No comments on this line yet.
+ 61 For reference: 150B pixels ≈ ~145,000 standard 1024×1024 images, which is a lot of image embeddings before you ever pay anything.
No comments on this line yet.
+ 62
No comments on this line yet.
+ 63 Source: Voyage AI pricing and Models Overview.
No comments on this line yet.
+ 64
No comments on this line yet.
+
65
No comments on this line yet.
+ 66
No comments on this line yet.
+ 68
No comments on this line yet.
+ 70
No comments on this line yet.
+ 71 ModelContextDimensionsNotes
No comments on this line yet.
+
72
voyage-4-large32,0001024 (default), 256 / 512 / 2048Best quality, multilingual
No comments on this line yet.
+
73
voyage-432,0001024 (default), 256 / 512 / 2048Balanced
No comments on this line yet.
+
74
voyage-4-lite32,0001024 (default), 256 / 512 / 2048Cheapest, lowest latency
No comments on this line yet.
+
75
voyage-context-332,0001024 (default), 256 / 512 / 2048Contextualized chunk embeddings
No comments on this line yet.
+
76
voyage-code-332,0001024 (default), 256 / 512 / 2048Code-tuned
No comments on this line yet.
+ 77
No comments on this line yet.
+
78
All embeddings produced by the voyage-4 series are interchangeable — you can mix voyage-4-large query vectors with voyage-4 document vectors in the same index.
No comments on this line yet.
+ 79
No comments on this line yet.
+ 81
No comments on this line yet.
+ 82 ModelContextDimensionsModalities
No comments on this line yet.
+
83
voyage-multimodal-3.532,0001024 (default), 256 / 512 / 2048Text + image + video
No comments on this line yet.
+
84
voyage-multimodal-3 (legacy)32,0001024Text + image
No comments on this line yet.
+ 85
No comments on this line yet.
+ 87
No comments on this line yet.
+ 88 ModelContextNotes
No comments on this line yet.
+
89
rerank-2.532,000Recommended for most apps
No comments on this line yet.
+
90
rerank-2.5-lite32,000Latency-optimized
No comments on this line yet.
+ 91
No comments on this line yet.
+
92
Legacy models still callable but not recommended: voyage-3-large, voyage-3.5, voyage-3.5-lite, voyage-code-2, voyage-multimodal-3, rerank-2, rerank-2-lite.
No comments on this line yet.
+ 93
No comments on this line yet.
+ 94 Source: Voyage AI Models Overview.
No comments on this line yet.
+ 95
No comments on this line yet.
+
96
No comments on this line yet.
+ 97
No comments on this line yet.
+ 99
No comments on this line yet.
+ 100 Free usage runs at Usage Tier 1. All limits are 2,000 RPM across the board; TPM varies by model:
No comments on this line yet.
+ 101
No comments on this line yet.
+ 102 ModelTPM (Tier 1)RPM (Tier 1)
No comments on this line yet.
+
103
voyage-4-lite, voyage-3.5-lite16,000,0002,000
No comments on this line yet.
+
104
voyage-4, voyage-3.58,000,0002,000
No comments on this line yet.
+
105
voyage-4-large3,000,0002,000
No comments on this line yet.
+
106
voyage-context-3, voyage-code-3, voyage-3-large, voyage-code-2, voyage-law-2, voyage-finance-23,000,0002,000
No comments on this line yet.
+
107
voyage-multimodal-3.5, voyage-multimodal-32,000,0002,000
No comments on this line yet.
+
108
rerank-2-lite, rerank-2.5-lite4,000,0002,000
No comments on this line yet.
+
109
rerank-2, rerank-2.52,000,0002,000
No comments on this line yet.
+ 110
No comments on this line yet.
+ 111 Users are automatically promoted to Tier 2 (2× limits) and Tier 3 (3× limits) based on billed (post-free) usage — you are never downgraded.
No comments on this line yet.
+ 112
No comments on this line yet.
+ 113 Source: Voyage AI Rate Limits.
No comments on this line yet.
+ 114
No comments on this line yet.
+
115
No comments on this line yet.
+ 116
No comments on this line yet.
+ 118
No comments on this line yet.
+ 119 Keep these in mind so you know what "normal" API spend looks like once you exceed free tokens:
No comments on this line yet.
+ 120
No comments on this line yet.
+ 121 ServicePrice
No comments on this line yet.
+
122
voyage-4-large$0.12 / 1M tokens
No comments on this line yet.
+
123
voyage-4$0.06 / 1M tokens
No comments on this line yet.
+
124
voyage-4-lite$0.02 / 1M tokens
No comments on this line yet.
+
125
voyage-context-3, voyage-code-3$0.18 / 1M tokens
No comments on this line yet.
+
126
rerank-2.5$0.05 / 1M tokens
No comments on this line yet.
+
127
rerank-2.5-lite$0.02 / 1M tokens
No comments on this line yet.
+ 128 Multimodal$0.12 / 1M text tokens + $0.60 / 1B pixels
No comments on this line yet.
+ 129 Batch (12-h SLA)33% discount
No comments on this line yet.
+ 130 File storage$0.05 / GB / month
No comments on this line yet.
+ 131
No comments on this line yet.
+
132
No comments on this line yet.
+ 133
No comments on this line yet.
+ 135
No comments on this line yet.
+ 136 • No auto-billing surprise: because no credit card is required, you cannot accidentally overrun into paid usage. Once free tokens for a given model are exhausted, calls to that model will start failing with a billing error until you add a payment method.
No comments on this line yet.
+
137
• Per-model bucket: the 200M is consumed per model — embedding the same documents with both voyage-4-large and voyage-4 to A/B them is free, but it does eat from each model's bucket independently.
No comments on this line yet.
+ 138 • Atlas account ties identity: the free tokens are scoped to your MongoDB Atlas Organization. Creating a second Org under the same email does not straightforwardly reset the bucket; MongoDB does enforce dedup signals.
No comments on this line yet.
+ 139 • Vector storage is on you: Voyage only returns vectors. If you want managed vector search, MongoDB Atlas Vector Search is the obvious pairing (free M0 cluster gives 512 MB which is enough for ~100k vectors at 1024 dims).
No comments on this line yet.
+ 140
No comments on this line yet.
+
141
No comments on this line yet.
+ 142
No comments on this line yet.
+ 144
No comments on this line yet.
+
145
• Use Matryoshka (MRL) dimensions to save cost: voyage-4-* supports 256 / 512 / 1024 / 2048 dims — you can request a smaller dimension and dramatically shrink your vector index without re-embedding.
No comments on this line yet.
+
146
• Reranker is cheap and underrated: rerank-2.5-lite at $0.02/M tokens (and 200M free) is one of the cheapest first-stage→second-stage rerank pipelines you can build today.
No comments on this line yet.
+ 147 • MongoDB integration: if you store your vectors in Atlas Vector Search, you can call Voyage embeddings directly from Atlas without managing your own embedding pipeline (see Atlas Vector Search docs).
No comments on this line yet.
+ 148 • Batch endpoint = 33% off for non-realtime workloads (12-h SLA) — useful for backfilling a large corpus.
No comments on this line yet.
+
149
• OpenAI-compatible-ish: the request/response shape is essentially OpenAI's /embeddings shape, so swapping in is a one-line change in most stacks (LangChain, LlamaIndex, Haystack all ship native Voyage integrations).
No comments on this line yet.
+ 150
No comments on this line yet.
+
151
No comments on this line yet.
+ 152
No comments on this line yet.
+ 153 Sources:
No comments on this line yet.
+ 154 • Voyage AI Pricing
No comments on this line yet.
+ 155 • Voyage AI Models Overview
No comments on this line yet.
+ 156 • Voyage AI Rate Limits
No comments on this line yet.
+ 157 • Voyage AI Quick Start
No comments on this line yet.
+ 158 • Model API Keys
No comments on this line yet.
+ 159 • Voyage AI FAQ
No comments on this line yet.
+ 160 • MongoDB blog: Embedding and Reranking API on Atlas
No comments on this line yet.