How to calculate token used from openAI while using Multimod LiveKit Community #ask-ai

Join Slack

How to calculate token used from openAI while usin...

# ask-ai

prehistoric-carpenter-68779

03/26/2025, 5:30 AM

How to calculate token used from openAI while using MultimodalAgent

tall-airline-16756

03/26/2025, 5:30 AM

Thanks for reaching out, Parth. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

tall-airline-16756

03/26/2025, 5:31 AM

To calculate the tokens used while using OpenAI's MultimodalAgent, you can refer to the following metrics: • Input Tokens: The number of tokens in the prompt, tracked as

promptTokens

. • Output Tokens: Includes both audio and text tokens, tracked as

audioTokens

and

textTokens

respectively. • Total Tokens: The sum of all tokens used, tracked as

totalTokens

. Additionally, the

ModelUsage

interface provides a detailed breakdown of input and output tokens, including: • `input_tokens`: Total input tokens • `input_token_details`: Breakdown of input tokens (audio, text, cached) • `output_tokens`: Total output tokens • `output_token_details`: Breakdown of output tokens (audio, text) • `total_tokens`: Sum of all tokens used Performance metrics like

tokensPerSecond

and

ttft

prehistoric-carpenter-68779

03/26/2025, 5:31 AM

can you provide the code exmaple of above

tall-airline-16756

03/26/2025, 5:31 AM

Unfortunately, I couldn't locate a specific code example for calculating tokens used by OpenAI's MultimodalAgent. However, you can use the metrics provided by the

MultimodalLLMMetrics

and

ModelUsage

interfaces to track and calculate token usage programmatically. If you have access to the implementation, you can retrieve values like

promptTokens

audioTokens

textTokens

, and

totalTokens

from these interfaces. If you need further assistance or a detailed example, please let me know, and I can escalate this to a human specialist for more support.

Open in Slack

Previous Next