Starter
Free
500 queries
Track Pipeline Performance
Search reranking
Email support
Pro
$60
/month
60k queries
Basic Alerting
Download Problematic Queries
Priority Support
Enterprise
Custom
Unlimited Custom Alerts
Fine-Tuned Models
Optional VPC Deployments
99.999% Uptime SLA
How does the semantic filter improve RAG outputs?
Our semantic filter technology maps the tokens from the query on to the provided corpus, this allows for greater accuracy than vector search alone. You can use Pongo in `observe` mode to quantify the improvement before modifying your pipeline.
Can I self host Pongo?
Yes, Pongo can be Self-hosted. Just book a call with us and we'll find the best option for you.
What is Pongo's latency?
Pongo adds 300ms to 400 ms for 50 documents of 512 tokens. We have deployments in Oregon and N. Virginia, please contact us if you need deployments in another region.
Is Pongo secure?
Yes, Pongo only stores data if you opt in. All data is encrypted, and cryptographically scrambled. No customer data is ever used to train our own models. We are in the process of getting SOC2 compliance.
Can I fine-tune Pongo?
Yes. If you fine-tune Pongo, the model will be your IP and you will own it. Contact us to get started.
Does Pongo use an LLM for reranking?
No Pongo uses our propriatery semantic filter technology, that is ~1000x faster than LLMs like GPT-4o and 4o mini, to rerank results at runtime. We use a large LLM in the background to analyze queries that are potentially missing context.
© Pongo Technologies Inc. 2024