spent 0.1.0

See what your AI really costs. Zero code changes.

spent

See what your AI really costs. Zero code changes.

---

You're spending hundreds on LLM APIs every month. Do you know which calls cost what?

spent tracks every token, every model, every dollar -- automatically. One command. No code changes. Beautiful reports.

$ spent run python app.py

 ┌──────────────────────────────────────────┐
 │  spent                    session a1b2c3 │
 │                                          │
 │  Total Cost:    $4.2731                  │
 │  Tokens:        125,430  (98k in / 27k out)
 │  API Calls:     47                       │
 │  Duration:      2m 34s                   │
 └──────────────────────────────────────────┘

  Model                   Calls  Tokens     Cost     Share
  gpt-4o                  12     84,200     $3.8100  ████████░░ 89%
  gpt-4o-mini             31     38,100     $0.4200  █░░░░░░░░░ 10%
  claude-sonnet-4-6        4      3,130     $0.0431  ░░░░░░░░░░  1%

  Savings Opportunities: ~$2.1000
    gpt-4o -> gpt-4o-mini: save $2.10 (55%) on 12 calls

Quick Start

pip install spent

Option 1: Zero code changes (recommended)

Just prefix your command:

spent run python app.py
spent run python -m pytest
spent run --budget 5.00 python train.py

Option 2: One line of code

from spent import track
from openai import OpenAI

client = track(OpenAI())

# Use normally -- costs tracked automatically
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}],
)
# Exit summary printed automatically

Works with Anthropic too:

from spent import track
from anthropic import Anthropic

client = track(Anthropic())

Features

Feature	Status
OpenAI cost tracking	Done
Anthropic cost tracking	Done
Beautiful terminal dashboard	Done
Per-model cost breakdown	Done
Savings recommendations	Done
Budget alerts	Done
Session history	Done
JSON/CSV export	Done
Google AI tracking	Roadmap
Auto model routing	Roadmap
Team cost sharing	Roadmap
CI/CD cost reports	Roadmap

Commands

# Track costs (zero code changes)
spent run python app.py

# Set a budget alert
spent run --budget 10.00 python app.py

# Live dashboard (watches costs in real-time)
spent dashboard

# View cost reports
spent report              # Recent sessions
spent report --today      # Today's costs
spent report --json       # Machine-readable
spent report --csv        # Spreadsheet-ready

# Clear all data
spent reset

How It Works

spent transparently patches the OpenAI and Anthropic Python SDKs at import time. When your code calls client.chat.completions.create(...), spent:

1. Intercepts the call (before and after)
2. Extracts token usage from the response
3. Calculates cost using up-to-date pricing
4. Stores the record in a local SQLite database
5. Returns the original response unchanged

Your code runs exactly the same. No API proxies. No network changes. No latency added to API calls themselves.

Data stays on your machine at ~/.spent/data.db.

Supported Models

40+ models with up-to-date pricing:

Provider	Models
OpenAI	GPT-4o, GPT-4o-mini, GPT-4-Turbo, GPT-4, GPT-3.5-Turbo, o1, o3, o3-mini, o4-mini
Anthropic	Claude Opus 4, Claude Sonnet 4, Claude Haiku 4, Claude 3.5/3 family
Google	Gemini 2.5 Pro/Flash, 2.0 Flash, 1.5 Pro/Flash
DeepSeek	DeepSeek Chat, DeepSeek Reasoner
Mistral	Mistral Large, Mistral Small, Codestral
Groq	Llama 3.3 70B, Llama 3.1 8B, Mixtral 8x7B

Unknown models are tracked with $0 cost (tokens still recorded).

Budget Alerts

Set a budget and get warned when you exceed it:

# CLI
spent run --budget 5.00 python app.py

# Python
from spent import track, configure
configure(budget=5.00)
client = track(OpenAI())

When the budget is exceeded:

[spent] BUDGET ALERT: $5.0231 spent (budget: $5.00)

Why spent?

	spent	Manual tracking	LLM framework built-in
Code changes needed	0	Lots	Framework-specific
Works across providers	Yes	Manual per-provider	Usually one provider
Historical data	SQLite	Spreadsheets	In-memory only
Savings recommendations	Automatic	You do the math	No
Export formats	JSON, CSV	Copy-paste	Varies
Privacy	100% local	Depends	Often cloud

Roadmap

• Google AI / Vertex tracking -- Gemini model support
• Auto model routing -- automatically use cheaper models for simple tasks
• Team dashboards -- aggregate costs across team members
• CI/CD integration -- cost reports in GitHub Actions / PR comments
• Ollama / local model tracking -- track local inference costs (compute time)
• Web dashboard -- browser-based cost explorer
• Slack / Discord alerts -- budget notifications in team channels
• Cost anomaly detection -- alert on unusual spending patterns

Contributing

Contributions welcome! See CONTRIBUTING.md for guidelines.

git clone https://github.com/loplop-h/spent.git
cd spent
pip install -e ".[dev]"
pytest

License

MIT