How to Track Your Total AI Spending Across Claude, OpenAI, Cursor, and Copilot

If you’re a developer in 2026, you’re probably paying for AI tools from at least two or three different providers. Each has its own billing dashboard, its own pricing model, and its own way of presenting usage data. The result is that most developers have no idea what they’re actually spending on AI in total. A 2025 Retool survey found that 64% of developers could not accurately estimate their combined monthly AI tool costs within 25% (Retool, “State of AI in Development,” 2025).

This guide covers how to get a clear picture of your total AI spending, which tools help with tracking, and how to build a monthly audit habit that keeps costs under control.

Why is tracking total AI spending so difficult?

Tracking total AI costs is difficult because developers use 3-4 AI tools simultaneously, each with different billing models — fixed subscriptions, per-token API charges, and usage-based tiers — spread across separate dashboards with no cross-platform aggregation. You need to visit 3-4 different billing pages, convert different units (tokens, credits, requests), and manually sum everything to get a total.

The modern developer AI stack typically looks something like this:

Tool	Billing Model	Where to Check	Update Frequency
Claude (API)	Per-token	console.anthropic.com	Every few hours
Claude (Pro/Max)	Monthly subscription	anthropic.com/account	Monthly
OpenAI (API)	Per-token	platform.openai.com/usage	Near real-time
ChatGPT Plus/Pro	Monthly subscription	openai.com/settings	Monthly
GitHub Copilot	Monthly subscription	github.com/settings/billing	Monthly
Cursor	Monthly subscription	cursor.com/settings	Monthly
Windsurf	Monthly subscription + credits	codeium.com/account	Monthly

That’s seven different places to check for a developer using a common tool combination. And the billing models are fundamentally different: subscriptions are fixed and predictable, API usage is variable and can spike without warning, and credit-based systems add yet another abstraction layer.

The fragmentation means that even disciplined developers lose track of their total spend. You know your Claude API bill was $85 last month. But when you add the $20 Copilot subscription, the $20 ChatGPT Plus, and the $20 Cursor Pro — suddenly you’re at $145/month on AI tools alone.

How can you track Claude and OpenAI costs locally?

FavTray monitors your Claude and OpenAI API usage directly from local log files on your Mac, displaying running cost totals in your menu bar without sending any data to external servers. It reads the ~/.claude/ directory for Claude usage and local API call logs for OpenAI, applying current per-token pricing to show session, daily, and weekly costs.

The local tracking approach solves two problems that billing dashboards don’t:

1. Real-time visibility: You see costs as they happen, not hours or days later
2. Per-session granularity: You know exactly which task or coding session drove which costs

For developers whose primary AI spending comes from Claude Code and OpenAI API calls, FavTray provides the majority of cost visibility directly in the macOS menu bar. Since these two providers typically account for 60-80% of a developer’s variable AI costs, having them tracked automatically covers the bulk of the spending that’s hardest to predict.

The remaining costs — fixed subscriptions to Copilot, Cursor, or ChatGPT Plus — are predictable monthly charges that don’t need real-time tracking. They need to be included in your total, but they don’t surprise you on the billing page.

What does a monthly AI cost audit look like?

A monthly AI cost audit takes 15 minutes and involves collecting costs from each provider, categorizing them as fixed versus variable, comparing against your budget, and deciding whether to adjust plans or cancel underused tools. Running this audit on the same day each month builds the habit that prevents spending from creeping up unnoticed.

Here’s a practical audit framework:

Step 1: Collect fixed costs List every AI subscription you pay for and its monthly cost:

• Claude Pro/Max: $20 / $100 / $200
• ChatGPT Plus/Pro: $20 / $200
• GitHub Copilot: $10 / $19
• Cursor Pro: $20
• Other tools: varies

Step 2: Collect variable costs Pull API spending from each provider’s billing dashboard:

• Anthropic console: Claude API usage for the month
• OpenAI platform: GPT API usage for the month
• If using FavTray, pull the monthly total from the app directly

Step 3: Calculate totals and trends

Category	This Month	Last Month	3-Month Average
Fixed subscriptions	$____	$____	$____
Claude API	$____	$____	$____
OpenAI API	$____	$____	$____
Total	$____	$____	$____

Step 4: Evaluate and adjust

• Any subscription unused in the past 2 weeks? Cancel it.
• API costs trending up? Implement token reduction techniques.
• Total exceeding your budget? Review which plan tier makes sense.

Developers who run this audit monthly typically reduce their AI spending by 15-25% over three months, primarily by canceling underused subscriptions and switching between plan tiers based on actual usage data.

How do you set a reasonable AI tool budget?

A reasonable AI tool budget for an individual developer ranges from $50 to $250 per month depending on usage intensity, with the 80th percentile of professional developers spending under $200/month across all tools. Start by setting your budget at 1-2% of your monthly income, then adjust based on the productivity return you’re getting.

Budget allocation frameworks that work in practice:

The Fixed-Plus-Variable approach:

• Allocate 40% of budget to fixed subscriptions (pick your 1-2 most-used tools)
• Allocate 60% to variable API usage
• Set daily API alerts at variable budget / 22 working days

The ROI-based approach:

• Estimate hours saved per week by AI tools
• Multiply by your effective hourly rate
• Your AI budget should be less than 30% of the value generated
• Example: AI saves 8 hours/week at $75/hour = $2,400/month value. Budget of $200/month = 8% of value generated.

The cap-and-review approach:

• Set a hard monthly cap you’re comfortable with
• Distribute across tools proportionally to usage
• Review quarterly and adjust caps up or down

The key insight is that most developers don’t overspend on AI tools because they use them too much — they overspend because they subscribe to tools they use infrequently. The median developer in 2025 had at least one AI subscription they used fewer than 3 times per month (Gartner, “Developer Tooling Spend Analysis,” 2025).

What tools help consolidate AI spending visibility?

For API-based costs (Claude and OpenAI), FavTray provides real-time local tracking on macOS. For subscription management, general-purpose tools like Copilot for Finance or even a simple spreadsheet work well. No single tool yet covers all AI providers in one dashboard, but combining local API tracking with a subscription ledger gets you 90% of the visibility you need.

Here’s a practical tracking stack:

Spending Type	Tracking Tool	Effort
Claude API costs	FavTray (local, real-time)	Automatic
OpenAI API costs	FavTray (local, real-time)	Automatic
Fixed subscriptions	Spreadsheet or budgeting app	5 min/month
Usage patterns	Editor time tracking + session logs	Varies

The missing piece in the ecosystem is a unified dashboard that pulls from every provider’s billing API. Until that exists, the combination of automated tracking for your heaviest variable costs (Claude and OpenAI via FavTray) plus a simple monthly subscription tally is the most practical approach.

Some developers have built custom scripts that pull billing data from each provider’s API and aggregate it into a personal dashboard. This works but requires maintenance as providers update their APIs. For most people, the 15-minute monthly audit described above is simpler and nearly as effective.

How does AI tool spending compare to other developer costs?

AI tools have become the second-largest category of developer tool spending after IDE and cloud infrastructure costs, averaging $100-200/month compared to $50-150 for traditional SaaS developer tools. For many developers, AI tool costs now exceed what they spend on hosting, databases, and monitoring combined.

Industry spending benchmarks for context:

Developer Tool Category	Typical Monthly Cost	Trend
Cloud infrastructure (personal/side projects)	$20 - $100	Stable
IDE and editor tools	$0 - $25	Stable
AI coding assistants (total)	$75 - $250	Rising 30-40% YoY
SaaS dev tools (monitoring, CI, etc.)	$30 - $80	Stable
Domain and hosting	$10 - $30	Stable

The year-over-year increase in AI tool spending is notable. According to Andreessen Horowitz, developer AI spending grew 35% between 2024 and 2025, and early 2026 data suggests continued acceleration as models become more capable and developers integrate them more deeply into workflows (a16z, “State of AI in Enterprise,” 2025).

This growth makes tracking and budgeting more important over time, not less. A cost that grows 35% per year unmonitored will double in just over two years. Building visibility and control now prevents a slow creep that becomes painful later.