Tokey is an LLM inference simulator — a tool for experiencing what different hardware and network configurations actually feel like when streaming tokens. You can compare a Raspberry Pi against an H100, a cloud API against a local GPU, or a reasoning model's long TTFT against a snappy consumer card, all without touching real hardware.
This repository is the issue tracker for the platform. There is no code here. If something is broken or you have an idea, open an issue and use one of the templates — they keep reports consistent and make triage faster.
The samples/ directory contains the simulator's built-in configurations as plain YAML, in case you want to reference them, fork them, or understand what drives a particular preset.
samples/presets/ — The 14 named presets available in the UI, covering GPUs (RTX 3070 through H200), cloud APIs, CPU inference (Pi 5, i7-13700K, Ryzen 9 9950X3D, Threadripper Pro), and model-type presets (MoE, Reasoning, Multimodal, Dense). Each file contains the full SimulationConfig: tokens per second, TTFT, prefill throughput, jitter, prompt and completion length, cadence batching, and stall settings.
samples/harnesses/ — The five content harnesses used to drive the simulator: code generation (a multi-turn Go task engine implementation), creative writing (a Paris short story), emails & business communication, conversational chat (a TypeScript walkthrough), and the custom harness placeholder. Each file contains the full multi-turn stage content — prompts, thinking text, and response text — exactly as used in the app.
Do not file security vulnerabilities as public issues. See SECURITY.md for responsible disclosure.