What Happened
A new tool called BurnRate launched on March 7, 2026, offering cost analytics across seven AI coding platforms: Claude Code, Cursor IDE, OpenAI Codex, GitHub Copilot, Windsurf, Cline, and Aider. It ships as a single binary for macOS, Linux, and Windows with no dependencies required.
The tool works locally. It parses session data from your AI tool directories, analyzes token usage patterns, and generates cost breakdowns without uploading any code or session content to external servers. It tracks over 40 AI models across the supported providers.
Key features include per-session cost breakdowns with API-equivalent pricing, provider comparison dashboards, project-level cost tracking, operation-level token analysis (separating reading, writing, searching, and thinking), rate limit monitoring, and 20 optimization rules with estimated savings.
Pricing starts free with basic session and cost views limited to 30-day history. The Pro tier at $12/month adds optimization rules, budget alerts, provider comparisons, and model recommendations. Team pricing runs $29/seat/month with aggregate dashboards and Slack alerts.
The creator built it after hitting the 5-hour rate limit on Claude Code Pro repeatedly without understanding which sessions were consuming the allocation or whether Opus was worth the premium over Sonnet for specific workflows.
Why It Matters
If you use multiple AI coding tools, you probably do not know what you are actually spending. Between Claude Code subscriptions, Cursor Pro, Copilot, and API usage, monthly costs can easily exceed $200 without a clear picture of the return.
BurnRate addresses a problem that most developers have been solving with gut feel: is this tool worth the subscription? Am I using it enough to justify the cost? Should I use a cheaper model for routine tasks and save the premium models for complex work?
The operation-level breakdown is particularly useful. Knowing how much of your token budget goes to "thinking" versus actual code generation helps you decide whether to adjust model settings or switch tools for certain task types. If 60% of your Claude Code tokens are consumed by reasoning on simple tasks, you might be overpaying.
The rate limit monitoring also fills a real gap. Claude Code's 5-hour rolling window is opaque by design, and tools like Cursor have similar allocation systems that are hard to track manually.
Our Take
This solves a genuine problem. AI coding tool costs are the new cloud bill - easy to ignore, hard to optimize, and constantly growing. Having a single dashboard across providers is valuable, especially as most developers now use two or three AI coding tools depending on the task.
The local-first approach is the right call. Nobody wants their session data and code patterns uploaded to a third-party service. The fact that BurnRate parses local session files means you get analytics without the privacy tradeoff.
The $12/month Pro tier is reasonable if you are spending $100+ on AI coding tools, but the free tier may be enough for individual developers who just want basic visibility. The team tier at $29/seat feels steep unless you are managing budgets across an engineering org.
One concern: this tool depends on parsing session data formats that the AI coding tool vendors control. If Claude Code or Cursor changes their local data format, BurnRate breaks until it is updated. That is an inherent fragility in any tool that reverse-engineers another product's local storage.
If you are running Claude Code Pro and have ever wondered where your 5-hour allocation actually goes, BurnRate is worth trying. Start with the free tier and see if the cost visibility changes how you choose models and allocate work across your AI tools.