Cost Optimizer

> Built by GetAgentIQ — [getagentiq.ai](https://getagentiq.ai)

Track, analyze, and optimize your OpenClaw API costs. Stop overpaying for simple tasks.

Quick Start

# Analyze current session costs
python3 scripts/cost-optimizer.py status

# Estimate cost for a task before running it
python3 scripts/cost-optimizer.py estimate --task "summarize a 2000-word article" --model claude-sonnet-4-20250514

# Get model recommendation for a task type
python3 scripts/cost-optimizer.py recommend --task-type simple

# Show cost comparison across models
python3 scripts/cost-optimizer.py compare

# Project monthly costs from current usage
python3 scripts/cost-optimizer.py project --days 7

# Set a budget alert
python3 scripts/cost-optimizer.py budget --daily 5.00 --monthly 150.00

How It Works

1. Track — Log token counts and model used per task

2. Analyze — Categorize tasks by complexity and map to optimal models

3. Recommend — Suggest the cheapest model that meets quality requirements

4. Alert — Warn before budget limits are hit

Task Complexity Classification

| Complexity | Examples | Recommended Model |

|------------|----------|-------------------|

| Simple | Formatting, search, short replies, reactions | Haiku / GPT-4o-mini / Gemini Flash |

| Medium | Summarization, code review, analysis | Sonnet / GPT-4o / Gemini Pro |

| Complex | Architecture, security audit, strategic planning | Opus / o1 / Gemini Ultra |

| Trivial | Heartbeats, status checks, file reads | Haiku / local Ollama |

Model Pricing (March 2026)

See references/model-pricing.md for full pricing table with input/output token rates.

Budget Management

Track spend against configurable limits. Integrates with SOUL.md budget caps if present.

# Check budget status
python3 scripts/cost-optimizer.py budget --status

# Output:
# Daily:   $2.34 / $5.00 (46.8%) ██████████░░░░░░░░░░
# Monthly: $47.20 / $180.00 (26.2%) █████░░░░░░░░░░░░░░
# Trend:   $4.71/day avg → projected $141.30/month ✅ Under budget

Parameters

| Command | Description |

|---------|-------------|

| status | Show current session/daily/monthly spend |

| estimate | Estimate cost for a described task |

| recommend | Get cheapest suitable model for task type |

| compare | Side-by-side model cost comparison |

| project | Project monthly cost from recent usage |

| budget | Set or check budget limits |

| log | Log a completed task's token usage |

| report | Generate cost report (text/json/markdown) |

Integration

• Reads SOUL.md for existing budget caps
• Works alongside claw-api-manager (internal budget enforcement)
• JSON output for cron-based monitoring
• Can recommend model switches mid-conversation