Skip to main content
The model is the brain of your Agent. Choosing the right one—and configuring it properly—determines how well your Agent reasons, how fast it responds, and how much it costs to run. Model configuration

​
Selecting a model

Cargo supports leading LLM providers:
ProviderModelsStrengths
OpenAIGPT-4o, GPT-4o-miniExcellent reasoning, broad capabilities, reliable
AnthropicClaude 3.5 Sonnet, Claude 3 OpusStrong reasoning, nuanced responses, good at following instructions
GoogleGemini Pro, Gemini FlashFast, cost-effective, good for simpler tasks

​
How to choose

Task complexityRecommended approach
Complex analysis (qualification, research synthesis)Use a powerful model (GPT-4o, Claude Opus)
Simple classification (routing, tagging)Use a faster model (GPT-4o-mini, Gemini Flash)
High volume, low stakesPrioritize speed and cost
Low volume, high stakesPrioritize reasoning quality
Start with a powerful model during development to understand what’s possible, then optimize for cost/speed once the Agent is working well.

​
Behavioral parameters

​
Reasoning Steps

The maximum number of logical sub-tasks the Agent can perform to reach a conclusion.
SettingEffectUse when
Low (1-3)Fast, direct responsesSimple lookups, single-tool tasks
Medium (4-6)Balanced reasoningMost standard workflows
High (7+)Deep, multi-step problem solvingComplex research, multi-source synthesis
Higher reasoning steps increase latency and cost. Only increase if the Agent is failing to complete tasks.

​
Temperature

Controls the creativity and variability of outputs.
TemperatureBehaviorBest for
0.0 - 0.2Deterministic, precise, consistentData extraction, classification, CRM updates
0.3 - 0.6Balanced creativityGeneral tasks, qualification, research
0.7 - 1.0Creative, variedCopywriting, brainstorming, conversational engagement
Rule of thumb:
  • Lower temperature → reproducible results
  • Higher temperature → more varied, human-like responses

​
Configuration examples

Lead Qualification Agent

Model: GPT-4o
Reasoning Steps: 4
Temperature: 0.2Needs good reasoning but consistent, reliable outputs.

Account Research Agent

Model: Claude 3.5 Sonnet
Reasoning Steps: 6
Temperature: 0.4Requires deeper research and synthesis across sources.

Email Drafting Agent

Model: GPT-4o
Reasoning Steps: 3
Temperature: 0.7Needs creativity for engaging, personalized copy.

Simple Router Agent

Model: GPT-4o-mini
Reasoning Steps: 2
Temperature: 0.1Fast classification with minimal reasoning needed.

​
Optimizing for cost and speed

Once your Agent works correctly, consider:
  1. Reduce reasoning steps if tasks complete in fewer steps
  2. Try a faster model (like GPT-4o-mini) and verify quality remains acceptable
  3. Lower temperature for more predictable outputs
  4. Limit resources to reduce context size and processing time