'%20stroke-width='4'%20stroke-linecap='round'/%3e%3cpath%20d='M24%2026%2031.1%2018.9'%20stroke='%23FF6B4A'%20stroke-width='3.5'%20stroke-linecap='round'/%3e%3ccircle%20cx='24'%20cy='26'%20r='2.8'%20fill='%23FF6B4A'/%3e%3c/svg%3e)
Questions before you install
Short, exact answers about cost, safety and what Tokrate shares. Everything here matches the privacy policy and methodology.
Cost and safety
Does Tokrate use my tokens or cost me anything?
No. Tokrate never calls a model and never sends prompts on your behalf. It reads the session logs that Codex, Claude Code and Grok Build already write on your computer and calculates speed from the token counts and timestamps in them, so it adds nothing to your token usage, API bill or subscription limits.
It does not ask for API keys and needs no Tokrate account. The companions and the public board are free, with no ads.
Is Tokrate safe to install?
The Mac app is signed with an Apple Developer ID and notarized by Apple. Windows and Linux builds are early alpha releases: the Windows installer is not Authenticode-signed yet, so Windows may show an unknown-publisher warning. Every release publishes SHA-256 checksums, and in-app updates are signature-verified before you choose to install them.
The companions are open source under the MIT license and contain no advertising or behavioral analytics SDKs. You can read exactly what leaves your computer, or ask your AI agent to review the code for you.
Does Tokrate slow down my coding agent?
No. Tokrate does not sit between your agent and the model, proxy any traffic or change your tool's settings. It reads new lines from the local session logs in small, bounded batches after they are written, so your agent runs exactly as it would without it.
Is Tokrate free?
Yes. The public board at tokrate.dev, the companions for Mac, Windows and Linux, and the public API are free to use, with no account, no ads and no paid tier.
Your data and code
Is my code, prompt or data shared with anyone?
No prompts, responses, code, file paths, session IDs, names or account details ever leave your computer. Sharing is optional: from version 0.1.11 it stays off until you choose “Yes, let's contribute”, and with “Only for local use” no measurements are sent at all (software update checks are a separate switch).
If you do share, each finished turn sends only performance numbers: coding tool, model, inference provider, recorded reasoning effort, client versions, token counts, durations, a five-minute time bucket and a random sample ID. The server stores your continent, never your country or IP address. Contribution data is not sold and is not used for advertising.
Does Tokrate read my code?
To measure speed, the companion reads your coding agent's local session logs, and those logs can contain your prompts and code. It extracts only timestamps, token counts and model details, keeps seven days of those numbers locally, and never copies or uploads the logs themselves.
Who can see what I contribute?
The public board shows only aggregates per model, and contributor identifiers are never published. Normally a number is published only after at least 10 contributors and 50 eligible turns; during launch, a clearly labelled Early data mode can show a result from a single installation, so it may reflect one person's workload.
Authorized operators at ILO APPLICATIONS SL, the company behind Tokrate, can access operational data. Cloudflare and the hosting provider process connection data such as IP addresses to deliver and protect the service.
Is sharing anonymous?
It is pseudonymous, not perfectly anonymous. Your installation signs contributions with a random key kept in your operating system's credential store, so your reports can be linked to each other but not to your name or accounts. Samples are stored with a hash of that key, not with your IP address.
How do I stop sharing or delete my data?
Switch sharing off at any time in the companion. That stops all contributions and community requests immediately, clears anything queued, and local monitoring keeps working. Server samples expire automatically after 30 days. For an earlier erasure request, email [email protected].
Measurements
What is response speed?
Response speed is output tokens per second while the model is responding: from the request to the finished answer, including the wait for the first token and the model's thinking tokens, excluding tool runs and your own time. Only responses with at least 200 output tokens count. It is Tokrate's headline number because it compares models fairly even when their tasks need different amounts of tool work.
How does Tokrate decide which model is fastest?
The board ranks models by the median response speed across contributors over the selected window, giving every installation equal weight so one heavy user cannot dominate. Models, inference providers, reasoning effort and coding tools stay separate instead of being blended into one number.
Why do Tokrate numbers differ from benchmarks or my own speed?
Tokrate observes real coding sessions, not a controlled benchmark. Task size, context length, reasoning effort, speed tier, provider and network distance all change a turn's speed, so treat the board as a community observation rather than a guarantee for your next task.
Does Tokrate measure answer quality?
No. A faster model is not a better one. Tokrate measures speed only; judge correctness and quality for your own work.
Which coding tools and platforms are supported?
Codex, Claude Code (local sessions, including subagents) and Grok Build. Response speed is available for Codex and Claude Code from companion 0.1.14 and for Grok Build from 0.1.15, where it is averaged over each turn's model calls. The Mac app is stable on macOS 14 or later with Apple silicon, and Windows and Linux are available as alpha builds.
More detail: the privacy policy, the methodology, and the download page.