TOKENEYES
- Your API key never leaves your browser. Memory only — gone when you close the tab.
- Direct browser → LLM calls. No TokenEyes server in the API path.
- Nothing logged or stored on our end.
- Use a free restricted key where available — providers have trial/free options.
- LLM providers may log requests per their own policies.
- Fun project disclaimer: model responses are generated by third-party AI systems. This interface only sends your prompt/image and displays returned output.
- No guarantees: outputs can be wrong, biased, or unexpected. Responsibility for model behavior belongs to the AI provider.
Access mode
Vision provider
Vision model
Perspective style
Normal: quips are generated alongside the price scan — same call, no extra cost.
Simple: skips quip generation entirely (slightly faster, saves a few tokens).
Token split
Analyzing…
TOKENS
—
All models breakdown ▾
| Model | Input | Output | Reasoning | Total | ~Req |
|---|
Split adjustable in settings. Defaults: 30% input / 20% reasoning / 50% output. Real usage varies — chat skews input-heavy; agentic/code work skews output & reasoning.