Your files stay on your device

# AI tools

Count tokens, check context fit, estimate API costs, format fine-tuning files, and read chat exports. Exact where an OpenAI tokenizer exists, labeled estimates where one doesn't. Everything runs in your browser; exports never leave the tab.

9 tools

- [Tokenizer playground

Count and inspect OpenAI tokens (o200k, cl100k, p50k, r50k) in a worker.
](/ai/tokenizer)
- [Context fit

Does your prompt fit? Exact OpenAI counts; labeled estimates for any other model.
](/ai/context-fit)
- [API cost estimator

What a prompt costs to run: dated price data, exact arithmetic, tiers applied.
](/ai/cost-estimator)
- [Fine-tune JSONL formatter

Validate and format training files: target presets, per-line errors, loss ledger.
](/ai/jsonl-formatter)
- [Chat export viewer

Read a ChatGPT export or JSONL log. Virtualized, local, nothing sent anywhere.
](/ai/chat-export)
- [Chat template renderer

Render a transcript to ChatML, Llama, Mistral, or Gemma format, with token counts.
](/ai/chat-template)
- [Model comparison

The dated model registry as a table: context, output ceiling, rates, long tiers, tokenizer status.
](/ai/model-comparison)
- [Vision token cost

Image tokens by documented tile/area rules, costed exactly. Versioned formulas, custom rates, no vendor calls.
](/ai/vision-cost)
- [Prompt diff

Before/after prompts with exact token deltas, a line diff, and the cost change per model.
](/ai/prompt-diff)

Your files are processed on your device. The few tools that talk to the internet say so before they do.

---

Canonical HTML: https://nutter.tools/ai/
Markdown version: https://nutter.tools/ai/index.md
Plain-text version: https://nutter.tools/ai/index.txt
Agent index: https://nutter.tools/llms.txt