Skip to main content
Use output_model to delegate final response formatting to a different model. The primary model handles reasoning and tool calls, then the output model formats the result.

​
Why Use an Output Model?

Use CaseExample
Better writingResearch with GPT-5.2, write with Claude Opus 4.5
Cost optimizationReason with DeepSeek, format with GPT-5-mini
Structured outputUse a model without native support, format with one that has it

​
Better Writing

GPT-5.2 excels at research and tool use, but Claude Opus 4.5 produces better prose. Combine them: 
from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.models.openai import OpenAIResponses
from agno.tools.hackernews import HackerNewsTools

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),       # Research and tool calls
    output_model=Claude(id="claude-opus-4-5"), # Creative writing
    output_model_prompt="Write an engaging, well-structured article based on these findings.",
    tools=[HackerNewsTools()],
)

agent.print_response("Write an article about the latest AI breakthroughs", stream=True)
The primary model gathers information from HackerNews. Claude Opus 4.5 transforms those findings into polished prose.

​
Cost Optimization

Use a capable but expensive model for complex reasoning, a cheaper model for formatting: 
from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.yfinance import YFinanceTools

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),       # Expensive: complex analysis + tools
    output_model=OpenAIResponses(id="gpt-5-mini"),  # Cheap: just formatting
    output_model_prompt="Summarize the analysis in 3 bullet points.",
    tools=[YFinanceTools()],
)

agent.print_response("Deep analysis of NVDA financials", stream=True)
Or use a cheaper reasoning model with a better formatting model: 
from agno.agent import Agent
from agno.models.deepseek import DeepSeek
from agno.models.openai import OpenAIResponses
from agno.tools.yfinance import YFinanceTools

agent = Agent(
    model=DeepSeek(id="deepseek-chat"),        # Cheap: reasoning + tools
    output_model=OpenAIResponses(id="gpt-5.2"), # Better formatting
    tools=[YFinanceTools()],
)

agent.print_response("Analyze AAPL stock performance", stream=True)

​
Structured Output Support

Some models lack native structured output. Use an output model that supports it: 
from pydantic import BaseModel, Field
from agno.agent import Agent
from agno.models.openai import OpenAIResponses
from agno.tools.hackernews import HackerNewsTools

class ArticleSummary(BaseModel):
    title: str
    key_points: list[str] = Field(description="3-5 main takeaways")
    sentiment: str = Field(description="positive, negative, or neutral")

agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),            # Primary reasoning
    output_model=OpenAIResponses(id="gpt-5.2"),    # Structured output
    output_schema=ArticleSummary,
    tools=[HackerNewsTools()],
)

response = agent.run("Summarize the top AI story on HackerNews")
summary: ArticleSummary = response.content
print(summary.key_points)

​
How It Works

  1. Primary model processes the request and executes tool calls
  2. Primary model generates an intermediate response
  3. Intermediate response passes to the output model
  4. Output model formats and returns the final response

​
Custom Formatting Prompt

Use output_model_prompt to control the output style: 
from agno.agent import Agent
from agno.models.anthropic import Claude
from agno.models.openai import OpenAIResponses
from agno.tools.hackernews import HackerNewsTools

# Executive summary style
agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    output_model=Claude(id="claude-sonnet-4-5"),
    output_model_prompt="Format as a concise executive summary. No fluff, just insights.",
    tools=[HackerNewsTools()],
)

# Technical documentation style
agent = Agent(
    model=OpenAIResponses(id="gpt-5.2"),
    output_model=Claude(id="claude-sonnet-4-5"),
    output_model_prompt="Format as technical documentation with code examples where relevant.",
    tools=[HackerNewsTools()],
)

​
When to Use Output Model

ScenarioRecommendation
Need better prose qualityoutput_model with Claude Opus 4.5
Need to reduce costsoutput_model with GPT-5-mini
Primary model lacks structured outputoutput_model + output_schema
Need specific formatting styleoutput_model + output_model_prompt
Simple structured data extractionoutput_schema alone (no output model needed)