AI Model Selection & Cost Optimization

OpsHub’s Agent System features intelligent multi-provider AI with automatic model selection between Anthropic Claude and OpenAI GPT-4. The system optimizes for quality, cost, and speed while maintaining transparency in model usage and costs.

Claude Sonnet 4.5 is the preferred model for financial services operations due to its superior reasoning capabilities, larger context window (200K tokens), and excellent cost-performance balance.

Supported AI Models

OpsHub supports 13+ AI models across two providers, giving you flexibility to choose the right model for each task.

Anthropic Claude (Preferred for Financial Services)

Claude Sonnet 4.5

Best for: Complex financial analysis, compliance checks, NAV validation

Cost: $0.003 input /$ 0.015 output per 1K tokens
Context: 200,000 tokens
Speed: Fast
Capabilities: Advanced reasoning, coding, analysis, function-calling

⭐ Recommended for all financial operations tasks

Claude 3.5 Sonnet

Best for: General operations, dashboard creation, data analysis

Cost: $0.003 input /$ 0.015 output per 1K tokens
Context: 200,000 tokens
Speed: Very Fast
Capabilities: Strong reasoning, coding, analysis

Claude 3 Opus

Best for: Most complex analysis requiring deepest reasoning

Cost: $0.015 input /$ 0.075 output per 1K tokens
Context: 200,000 tokens
Speed: Moderate
Capabilities: Maximum reasoning depth

Claude 3 Haiku

Best for: Simple queries, data lookups, quick tasks

Cost: $0.00025 input /$ 0.00125 output per 1K tokens
Context: 200,000 tokens
Speed: Very Fast
Capabilities: Fast responses, simple tasks

OpenAI Models (Fallback & Alternative)

GPT-4o

Best for: Complex tasks when Claude unavailable

Cost: $0.0025 input /$ 0.01 output per 1K tokens
Context: 128,000 tokens
Speed: Fast
Capabilities: Strong reasoning, coding, analysis

GPT-4o Mini

Best for: Simple tasks, cost optimization

Cost: $0.00015 input /$ 0.0006 output per 1K tokens
Context: 128,000 tokens
Speed: Very Fast
Capabilities: Fast, simple tasks

O1

Best for: Advanced reasoning, complex problem-solving

Cost: $0.015 input /$ 0.06 output per 1K tokens
Context: 128,000 tokens
Speed: Moderate
Capabilities: Deep reasoning, complex analysis

O1 Mini

Best for: Moderate reasoning tasks

Cost: $0.003 input /$ 0.012 output per 1K tokens
Context: 128,000 tokens
Speed: Fast
Capabilities: Good reasoning, balanced cost

Intelligent Auto-Selection

When you don’t specify a model, OpsHub’s intelligent routing system automatically selects the best model based on:

Task Complexity Analysis

Analyzes message length, required tools, and context size to determine task complexity

Provider Availability

Checks which AI providers are configured and available

Financial Services Optimization

Prioritizes Claude Sonnet 4.5 for its superior performance on financial operations tasks

Cost-Performance Balance

Selects models that balance quality with cost-effectiveness

Auto-Selection Priority

Default Selection Order:

Claude Sonnet 4.5 (if Anthropic configured) - Preferred for financial services
Claude 3.5 Sonnet (fallback) - Excellent general performance
GPT-4o (if only OpenAI configured) - Strong alternative
GPT-4o Mini (simple tasks, OpenAI only) - Cost optimization

Manual Model Selection

You can manually select any available model through the Agent Console settings:

Via UI
Via API

Open the Agent Console
Click the Model Settings icon (⚙️) in the chat header
Choose your preferred model from the dropdown
Toggle between Auto (intelligent routing) or select a specific model
Set cost limits and speed preferences

Cost Tracking & Optimization

Real-Time Cost Visibility

Every agent interaction displays:

Model Used

See which AI model was selected for each response

Token Usage

Input and output tokens consumed

Estimated Cost

Real-time cost calculation in USD

Cost per Request

Stream completion events include full cost breakdown:

{
  "type": "done",
  "summary": {
    "modelProvider": "anthropic",
    "modelName": "claude-sonnet-4-5-20250514",
    "textChunks": 5,
    "toolCalls": 2,
    "inputTokens": 1234,
    "outputTokens": 567,
    "estimatedCost": 0.0123,
    "durationMs": 1234
  }
}

Monthly Cost Monitoring

Access the Monitoring Dashboard to view:

Total monthly AI costs by provider
Cost breakdown by agent and task type
Most expensive queries for optimization
Cost trends over time

Cost Optimization Tips

Use Haiku for Simple Tasks

For simple data lookups or quick questions, Claude Haiku is 10x cheaper than Sonnet while still providing excellent quality.Example: “What is today’s NAV for Fund ABC?” → Use Haiku

Leverage Auto-Selection

Let the intelligent routing system choose models based on task complexity. It optimizes for cost-performance automatically.Benefit: 20-30% cost savings vs. always using premium models

Use Concise Prompts

Clear, concise prompts reduce token usage while improving response quality.Instead of: “I need you to help me analyze the NAV for all funds and show me any issues…” Try: “Analyze NAV for all funds, flag issues >3bp”

Leverage Context Awareness

Agents remember your workspace context. No need to repeat fund names, dates, or previously mentioned information.Saves: 100-200 tokens per request

Set Cost Limits

Configure maximum cost per request in Model Settings to prevent expensive operations.Example: Set $0.10 limit for routine queries

Why Claude for Financial Services?

Superior Financial Reasoning

Claude Sonnet 4.5 demonstrates exceptional performance on complex financial analysis, regulatory compliance, and multi-step calculations.

Larger Context Window

200K token context (vs. 128K for GPT-4o) handles larger portfolios, longer compliance documents, and complex fund structures.

Better Code Generation

Excels at generating financial formulas, SQL queries for complex fund structures, and spreadsheet automation.

Excellent Cost-Performance

Competitive pricing (

0.003/

0.015) with superior quality makes it ideal for high-volume operations tasks.

Strong Function Calling

Reliable tool execution for our 62+ enterprise tools, with better handling of complex multi-tool workflows.

Consistent Output

More consistent responses for compliance and audit requirements where repeatability is critical.

Fallback & Reliability

OpsHub’s multi-provider architecture ensures continuous availability even if one provider experiences issues.

Automatic Fallback

If your preferred provider is unavailable:

Detection

System detects provider unavailability (API error, rate limit, timeout)

Fallback Selection

Automatically routes to alternative provider (Claude → GPT-4o or vice versa)

User Notification

Agent message indicates fallback: “Using GPT-4o (Claude temporarily unavailable)”

Quality Maintained

Both providers support all agent capabilities, ensuring consistent functionality

Provider Health Monitoring

Check provider status via health endpoint:

curl https://api.opshub.ai/agent/health

{
  "status": "healthy",
  "openai_configured": true,
  "anthropic_configured": true,
  "default_provider": "anthropic"
}

Best Practices

Financial Analysis

Use: Claude Sonnet 4.5Best for NAV validation, attribution analysis, compliance checks, and complex calculations.

Simple Lookups

Use: Claude Haiku or GPT-4o MiniBest for “What is…” queries, data retrieval, and quick status checks.

Dashboard Creation

Use: Claude 3.5 Sonnet or GPT-4oGood balance of cost and capability for visualization tasks.

Complex Workflows

Use: Claude Sonnet 4.5 or O1Best for multi-step processes requiring deep reasoning.

Batch Operations

Use: Auto-selection with cost limitsLet intelligent routing optimize across large volumes.

Compliance Reports

Use: Claude Sonnet 4.5Consistency and accuracy critical for regulatory requirements.

API Reference

Model IDs

When specifying models via API, use these exact model IDs: Anthropic:

claude-sonnet-4-5-20250514 - Claude Sonnet 4.5 (latest)
claude-3-5-sonnet-20241022 - Claude 3.5 Sonnet
claude-3-opus-20240229 - Claude 3 Opus
claude-3-haiku-20240307 - Claude 3 Haiku

OpenAI:

gpt-4o - GPT-4o
gpt-4o-mini - GPT-4o Mini
gpt-4-turbo - GPT-4 Turbo
o1 - O1
o1-mini - O1 Mini

Auto-Selection:

auto - Intelligent model selection (recommended)

Example Usage

// Auto-selection (recommended)
modelId: "auto"

// Specific model for financial analysis
modelId: "claude-sonnet-4-5-20250514"

// Cost optimization for simple query
modelId: "claude-3-haiku-20240307"

FAQ

Can I change the default model?

Yes! In Model Settings, you can set your preferred default model. The system will use it for all requests unless you specify otherwise.

What happens if my API key is invalid?

The system will automatically fall back to the alternative provider. You’ll see a notification indicating the fallback occurred.

How accurate is the cost estimation?

Cost estimates are calculated in real-time based on actual token usage and current provider pricing. Accuracy is >99% for completed requests.

Can I set cost budgets per user or team?

Yes, contact your OpsHub administrator to configure per-user or per-team cost limits and budget alerts.

Do different models support different tools?

No, all supported models have access to the same 62+ enterprise tools. Model selection doesn’t affect available capabilities.

Why is Claude preferred over GPT-4?

Claude Sonnet 4.5 demonstrates superior performance on financial operations tasks, particularly for complex reasoning, large context handling, and consistent regulatory compliance outputs. It’s also competitively priced.

Questions about model selection? Contact support@opshub.ai or check the Architecture Documentation for technical details.

Agent System Overview

Available Agents

Integration Patterns

Agent Tools & Capabilities

​AI Model Selection & Cost Optimization

​Supported AI Models

​Anthropic Claude (Preferred for Financial Services)

Claude Sonnet 4.5

Claude 3.5 Sonnet

Claude 3 Opus

Claude 3 Haiku

​OpenAI Models (Fallback & Alternative)

GPT-4o

GPT-4o Mini

O1

O1 Mini

​Intelligent Auto-Selection

​Auto-Selection Priority

​Manual Model Selection

​Cost Tracking & Optimization

​Real-Time Cost Visibility

Model Used

Token Usage

Estimated Cost

​Cost per Request

​Monthly Cost Monitoring

​Cost Optimization Tips

​Why Claude for Financial Services?

Superior Financial Reasoning

Larger Context Window

Better Code Generation

Excellent Cost-Performance

Strong Function Calling

Consistent Output

​Fallback & Reliability

​Automatic Fallback

​Provider Health Monitoring

​Best Practices

Financial Analysis

Simple Lookups

Dashboard Creation

Complex Workflows

Batch Operations

Compliance Reports

​API Reference

​Model IDs

​Example Usage

​FAQ

AI Model Selection & Cost Optimization

Supported AI Models

Anthropic Claude (Preferred for Financial Services)

OpenAI Models (Fallback & Alternative)

Intelligent Auto-Selection

Auto-Selection Priority

Manual Model Selection

Cost Tracking & Optimization

Real-Time Cost Visibility

Cost per Request

Monthly Cost Monitoring

Cost Optimization Tips

Why Claude for Financial Services?

Fallback & Reliability

Automatic Fallback

Provider Health Monitoring

Best Practices

API Reference

Model IDs

Example Usage

FAQ