[FEATURE] Add LiteLLM Cost Tracking to Agent Results

[iyamabhishek]

Problem Statement

Summary

Add support for returning LiteLLM cost data as part of AgentResult metadata when using LiteLLMModel.

Motivation

Currently, Strands agents track token usage via accumulated_usage, but there's no direct way to get the dollar cost of model invocations. LiteLLM provides a completion_cost() function that calculates costs based on token usage and model pricing, but this isn't exposed through the Strands SDK.

Proposed Solution

Proposed Solution

Add a pass-through function in Strands' LiteLLM model provider (strands/models/litellm.py) that:

1. Transforms Strands agent result to LiteLLM response format
2. Calls litellm.completion_cost() to calculate the cost
3. Includes the cost in AgentResult metadata

Example Usage

from strands import Agent
from strands.models.litellm import LiteLLMModel

model = LiteLLMModel(
    model_id="anthropic/claude-3-7-sonnet-20250219"
)

agent = Agent(model=model)
response = agent("What is 2+2")

# Access cost from response metadata
print(f"Cost: ${response.metadata.cost:.10f}")

Implementation Notes

The completion_cost function in LiteLLM has provider-specific logic for each model, so this would leverage existing LiteLLM functionality rather than reimplementing cost calculation.

Use Case

Users need cost tracking for:

• Budget monitoring and cost optimization
• Per-request cost analysis
• Comparing costs across different models
• Production cost tracking and reporting

Alternatives Solutions

No response

Additional Context

No response