CascadeResult

Returned by CascadeAgent.run(), run_streaming(), and run_batch(). Contains the generated response along with full cost, quality, timing, and routing diagnostics.

Usage

result = await agent.run("Explain quantum computing")

print(result.content)
print(f"Model: {result.model_used}")
print(f"Cost: ${result.total_cost:.6f}")
print(f"Savings: {result.savings_percentage}%")
print(f"Cascaded: {result.cascaded}, Accepted: {result.draft_accepted}")

Core Fields

Field	Type	Description
`content`	`str`	Generated response text
`model_used`	`str`	Model that produced the response
`total_cost`	`float`	Total cost in USD
`latency_ms`	`float`	Total latency in milliseconds
`complexity`	`str`	Detected complexity level
`cascaded`	`bool`	Whether cascade was used
`draft_accepted`	`bool`	Whether the draft passed quality validation
`routing_strategy`	`str`	Routing strategy used (`"direct"` or `"cascade"`)
`reason`	`str`	Explanation for the routing decision

Tool Calling

Field	Type	Description
`tool_calls`	`list[dict] \| None`	Tool calls made during execution
`has_tool_calls`	`bool`	Whether the response includes tool calls

Quality Diagnostics

Field	Type	Description
`quality_score`	`float \| None`	Quality score (0-1)
`quality_threshold`	`float \| None`	Threshold used for validation
`quality_check_passed`	`bool \| None`	Whether the quality check passed
`rejection_reason`	`str \| None`	Why the draft was rejected

Response Tracking

Field	Type	Description
`draft_response`	`str \| None`	Full draft response text
`verifier_response`	`str \| None`	Full verifier response text
`response_length`	`int \| None`	Response character length
`response_word_count`	`int \| None`	Response word count

Timing Breakdown

Field	Type	Description
`complexity_detection_ms`	`float \| None`	Time to detect complexity
`draft_generation_ms`	`float \| None`	Draft model generation time
`quality_verification_ms`	`float \| None`	Quality validation time
`verifier_generation_ms`	`float \| None`	Verifier model generation time
`cascade_overhead_ms`	`float \| None`	Overhead from cascade (wasted if draft rejected)

Cost Breakdown

Field	Type	Description
`draft_cost`	`float \| None`	Cost of the draft call
`verifier_cost`	`float \| None`	Cost of the verifier call
`cost_saved`	`float \| None`	Savings vs always using best model
`savings_percentage`	`float \| None`	Savings as percentage (0-100)

Model Information

Field	Type	Description
`draft_model`	`str \| None`	Draft model name
`draft_latency_ms`	`float \| None`	Draft model latency
`draft_confidence`	`float \| None`	Draft confidence score
`verifier_model`	`str \| None`	Verifier model name
`verifier_latency_ms`	`float \| None`	Verifier model latency
`verifier_confidence`	`float \| None`	Verifier confidence score

Methods

`to_dict()`

Convert the result to a plain dictionary.

data = result.to_dict()
import json
print(json.dumps(data, indent=2))

Python

TypeScript

CascadeResult

CascadeResult

Usage

Core Fields

Tool Calling

Quality Diagnostics

Response Tracking

Timing Breakdown

Cost Breakdown

Model Information

Methods

`to_dict()`

​CascadeResult

​Usage

​Core Fields

​Tool Calling

​Quality Diagnostics

​Response Tracking

​Timing Breakdown

​Cost Breakdown

​Model Information

​Methods

​to_dict()

CascadeResult

Usage

Core Fields

Tool Calling

Quality Diagnostics

Response Tracking

Timing Breakdown

Cost Breakdown

Model Information

Methods

`to_dict()`