Get This Tool
AgentMeter
Pricing
- Model
- Free
Summary
Your API dashboard shows per-request token counts. It does not show you that a single Claude Code session burned $50 before lunch, or why prompt caching cut your bill by 40% last week but nothing this week. AgentMeter sits between your agent and that confusion.
AgentMeter runs locally — no cloud sync, no account creation, no vendor dashboard to log into — and parses the tool calls, token counts, and caching splits that CLI agents like Claude Code, Gemini CLI, Codex CLI, and Copilot CLI generate. It surfaces the three-tier cost structure that prompt caching creates (input, cached-input, and output tokens each priced differently), which the raw API bill flattens into noise. The value-multiplier calculation compares API spend against estimated developer time saved, giving you a number to put in front of a manager. The wall appears when you need alerting, real-time budget enforcement, or integration with a team billing system — none of that is here.
Bottom line: Pick this when one developer needs to understand where their Claude Code budget is actually going; plan something else when your team needs spend caps, shared dashboards, or alerts before the bill arrives.
Community Performance Report Card
No community ratings yet. Be the first to rate this tool!
Community Benchmarks Community
Sign in to submit a benchmarkNo community benchmarks yet. Be the first to share a real-world data point.
Pros
Sign in to edit- Runs entirely on-device with no account, no cloud sync, and no vendor access to your session data, so usage patterns and project names never leave your machine.
- Breaks prompt-caching costs into the three actual billing tiers (input, cached-input, output), so you can see whether your caching strategy is paying off instead of inferring it from a flattened total.
- Per-session and per-project cost aggregation across Claude Code, Gemini CLI, Codex CLI, and Copilot CLI, which means you get a unified spend view instead of hunting across four separate dashboards.
- Value-multiplier calculation compares API spend against estimated developer time saved, so you have a concrete number when someone asks whether the agent usage is worth the invoice.
- Open-source under Apache-2.0, so you can audit exactly what it reads and how costs are calculated — no black-box pricing assumptions you have to take on faith.
Cons
Sign in to edit- There are no budget caps or threshold alerts. A session can exhaust your API credits before AgentMeter reports on it — the tool tells you what happened after the fact, not while it is happening. Teams that need spend enforcement have to wire up separate controls at the API key or infrastructure level.
- No shared or multi-user view exists. If two developers are both running Claude Code on the same project, their session data stays on their own machines. Teams that need consolidated spend reporting across contributors cannot get it here and will move to a vendor-native dashboard or a shared cost-tracking layer instead.
- Support is limited to CLI agents (Claude Code, Gemini CLI, Codex CLI, Copilot CLI). If your stack includes API-direct integrations, LangChain pipelines, or custom agent frameworks, AgentMeter produces nothing — you are back to reading raw API logs.
Community Reviews
Sign in to write a reviewNo reviews yet. Be the first to share your experience.
About
- Platforms
- macOS, Linux, Windows (Python)
- API Available
- No
- Self-Hosted
- Yes
- Last Updated
- 2026-06-12T08:15:58.826Z
Best For
Who it's for
- Developers using Claude Code, Gemini CLI, Codex CLI, or Copilot CLI
- Users wanting local-only cost visibility without vendor dashboards
- Teams optimizing prompt caching strategies
What it does well
- Track per-session API costs for Claude Code and similar agents
- Analyze prompt caching performance and savings
- Project monthly spend and compare model tiers
- Calculate value multiplier of agent usage versus developer time
Integrations
Discussion Community
Sign in to commentNo discussion yet. Sign in to start the conversation.
Spotted incorrect or missing data? Join our community of contributors.
Sign Up to ContributeCommunity Notes & Tips Community
Sign in to contributeBe the first to contribute. General notes, observations, gotchas, and tips from people who use this tool day-to-day.
Frequently Asked Questions
- Is AgentMeter free?
- Yes — AgentMeter is fully free to use. There is no paid tier.
- Is AgentMeter open source?
- Yes. AgentMeter is open source.
- Can I self-host AgentMeter?
- Yes. AgentMeter supports self-hosting on your own infrastructure.
- What platforms does AgentMeter support?
- AgentMeter is available on: macOS, Linux, Windows (Python).
Hours Saved & ROI Stories Community
Sign in to contributeBe the first to contribute. Concrete time/cost savings, with context. e.g. "Cut my code review backlog from 4h to 45m per week."
Curated lists that include this category
Most AI coding agent cost data arrives too late and too aggregated to act on. AgentMeter captures every tool call a CLI agent makes, tallies real token costs split across the three prompt-caching tiers, and reports totals per session and per project — all from a local `pip install` with no external dependencies. The core workflow is a CLI command against your local session logs: `agentmeter cost ProjectX` returns total spend, token volume, and call count in one line.
The differentiating capability is prompt-cache analysis. The vendor states that prompt caching splits costs into three token types at three different rates — input, cached-input, and output — and that AgentMeter makes this split visible where API dashboards show only aggregate numbers. For teams iterating on large context windows or repeated code-review prompts, knowing the cache hit rate for a session is the difference between tuning toward savings and guessing.
AgentMeter fits a single developer or a small team that wants local-only visibility into CLI agent spend without touching a vendor product. It breaks down when the need shifts to enforcement: there are no budget caps, no alerts when a session crosses a threshold, and no multi-user shared view. The tool is passive — it reads and reports, it does not intercept or block. Teams that discover they need spend governance rather than spend visibility will outgrow it quickly and move toward infrastructure-level controls or vendor-native quota tools.
The tool is open-source under the Apache-2.0 license, self-hosted by definition (it runs on the user’s machine), and installable via pip. It targets Claude Code as the primary integration, with stated support for Gemini CLI, Codex CLI, and Copilot CLI. The README notes a value-multiplier calculation that estimates API cost against developer time saved — useful for justifying agent usage to stakeholders who see only the invoice line.