Turbopuffer vs Qdrant
A comparison of Turbopuffer and Qdrant built from values read directly from each provider's own documentation, with the source recorded against every figure.
The short answer
- Pick Turbopuffer if
- Teams that want vector search billed like storage rather than compute — object-storage-backed pricing tends to undercut memory-resident vector databases at rest, especially for large, cold datasets.
- Pick Qdrant if
- Teams that want a choice of similarity metric rather than one fixed default, or that want the option to self-host the same open-source engine the managed cloud runs.
2 sourced criteria separate them:
- Pricing model
- Turbopuffer: Subscription with a monthly minimum per tier, plus usage-based overage charges beyond that minimum.
- Qdrant: Usage-based on the Standard tier: billed for compute (vCPU), memory (GB) and storage (GB) consumed by clusters, backup storage, and inference tokens for paid models.
- Similarity metrics supported
- Turbopuffer: Cosine distance is documented as a supported metric (used in the SDK's example code); the full list of supported metrics wasn't confirmed beyond this.
- Qdrant: Dot product, Cosine similarity (implemented as dot-product over automatically normalized vectors), Euclidean distance, and Manhattan distance.
Pick the criteria you care about. The chart counts how many of them lean toward each provider — the same read as scanning the bars below, just totalled for the ones you chose.
Turbopuffer 0
Qdrant 0
At a glance.
| Criterion | Turbopuffer | Qdrant |
|---|---|---|
| Pricing | ||
| Pricing model | Subscription with a monthly minimum per tier, plus usage-based overage charges beyond that minimum. | Usage-based on the Standard tier: billed for compute (vCPU), memory (GB) and storage (GB) consumed by clusters, backup storage, and inference tokens for paid models. |
| Vector | ||
| Similarity metrics supported | Cosine distance is documented as a supported metric (used in the SDK's example code); the full list of supported metrics wasn't confirmed beyond this. | Dot product, Cosine similarity (implemented as dot-product over automatically normalized vectors), Euclidean distance, and Manhattan distance. |
Where they differ.
Pricing model
- Turbopuffer
- Subscription with a monthly minimum per tier, plus usage-based overage charges beyond that minimum.
- Qdrant
- Usage-based on the Standard tier: billed for compute (vCPU), memory (GB) and storage (GB) consumed by clusters, backup storage, and inference tokens for paid models.
Sources (2) →Sources ↓
- Turbopuffer Pricing ↗
“Minimum monthly commitments ($16 Launch, $256 Scale, $4,096+ Enterprise), plus usage-based overage charges.”
Read 2026-09-11 · official pricing
- Qdrant Cloud Pricing ↗
“Billing is calculated based on actual resource usage during the billing period. You're charged for compute (vCPU), memory (GB), storage (GB) consumed by your clusters, storage (GB) consumed by backups, and used inference tokens of paid models.”
Read 2026-09-11 · official pricing
Similarity metrics supported
- Turbopuffer
- Cosine distance is documented as a supported metric (used in the SDK's example code); the full list of supported metrics wasn't confirmed beyond this.
- Qdrant
- Dot product, Cosine similarity (implemented as dot-product over automatically normalized vectors), Euclidean distance, and Manhattan distance.
Sources (2) →Sources ↓
- Turbopuffer documentation ↗
“distance_metric="cosine_distance"”
Read 2026-09-11 · official docs
- Qdrant documentation ↗
“Dot product (Dot), Cosine similarity (Cosine), Euclidean distance (Euclid), Manhattan distance (Manhattan). Cosine similarity is implemented as dot-product over normalized vectors. Vectors are automatically normalized during upload.”
Read 2026-09-11 · official docs
When these numbers change, hear about it. Sources are re-checked monthly; a repricing goes out as a short note.
Which should you choose?
Choose Turbopuffer if…
Teams that want vector search billed like storage rather than compute — object-storage-backed pricing tends to undercut memory-resident vector databases at rest, especially for large, cold datasets.
Editorial · Palash Bagchi · approved
Choose Qdrant if…
Teams that want a choice of similarity metric rather than one fixed default, or that want the option to self-host the same open-source engine the managed cloud runs.
Editorial · Palash Bagchi · approved
Consider something else if…
- Turbopuffer: The full list of supported similarity metrics matters before committing — that wasn't fully documented in what's sourced here, unlike competitors that publish the complete list.
- Qdrant: A published, calculator-free entry price matters — Qdrant Cloud's Standard tier pricing is usage-based with no static per-month figure quoted, unlike some competitors' flat entry plans.
Documented by only one.
These criteria are published by one provider and not the other. An absence here means we have not found a source, not that the feature is missing.
Questions this comparison answers.
Should I choose Turbopuffer or Qdrant?
Pick Turbopuffer if Teams that want vector search billed like storage rather than compute — object-storage-backed pricing tends to undercut memory-resident vector databases at rest, especially for large, cold datasets.
Pick Qdrant if Teams that want a choice of similarity metric rather than one fixed default, or that want the option to self-host the same open-source engine the managed cloud runs.