Pricing and limits ↗

In this case, the August invoice would only show line items for each service you used (totaling $100). Since your usage exceeds the minimum usage commitment, you are only charged for your actual usage and no additional minimum usage line item appears on your invoice.

Hourly rate#

For paid plans, you are charged an hourly rate for each assistant, regardless of assistant activity.

Tokens#

For paid plans, you are charged for the number of tokens used by each assistant.

Chat tokens#

Chatting with an assistant involves both input and output tokens:

• Input tokens are based on the messages sent to the assistant and the context snippets retrieved from the assistant and sent to a model. Messages sent to the assistant can include messages from the chat history in addition to the newest message.

• Output tokens are based on the answer from the model.

Chat input tokens appear as “Assistants Input Tokens” on invoices and prompt_tokens in API responses. Chat output tokens appear as “Assistants Output Tokens”" on invoices and completion_tokens in API responses.

Context tokens#

When you retrieve context snippets, tokens are based on the messages sent to the assistant and the context snippets retrieved from the assistant. Messages sent to the assistant can include messages from the chat history in addition to the newest message.

Context retrieval tokens appear as Assistants Context Tokens Processed on invoices and prompt_tokens in API responses. In API responses, completion_tokens will always be 0 because, unlike for chat, there is no answer from a model.

Evaluation tokens#

Evaluating responses involves both input and output tokens:

• Input tokens are based on two requests to a model: The first request contains a question, answer, and ground truth answer, and the second request contains the same details plus generated facts returned by the model for the first request.
• Output tokens are based on two responses from a model: The first response contains generated facts, and the second response contains evaluation metrics.

Evaluation input tokens appear as Assistants Evaluation Tokens Processed on invoices and prompt_tokens in API responses. Evaluation output tokens appear as as Assistants Evaluation Tokens Out on invoices and completion_tokens in API responses.

Storage#

For paid plans, you are charged for the size of each assistant.

Limits#

Pinecone Assistant limits vary based on subscription plan.

Object limits#

Object limits are restrictions on the number or size of assistant-related objects.

Additionally, the following limits apply to multimodal PDFs (currently in public preview):

Rate limits#

Rate limits help protect your applications from misuse and maintain the health of our shared infrastructure. These limits are designed to support typical production workloads while ensuring reliable performance for all users.

Most rate limits can be adjusted upon request. If you need higher limits to scale your application, contact Support with details about your use case.

Requests that exceed a rate limit fail and return a 429 - TOO_MANY_REQUESTS status.

To handle rate limits, implement retry logic with exponential backoff.