LLM Cost Estimator
Estimate multi-model API cost from tokens or pasted prompts.
OpenScore prompt variants against weighted rubrics before spending API credits. - runs entirely in your browser. Free, fast and private.
Compare two or three candidates side by side.
Score clarity, constraints, few-shot, and output format.
Heuristic checks run privately with no model call.
Share results or save the last rubric in localStorage.
Prompt Eval Rubric is a free ai prompt toolkit utility on OneDevToolkit. Score prompt variants against weighted rubrics before spending API credits. It runs in your browser for this session - use the workspace on this page to try it.
Need a related step? Open the AI Prompt Toolkit hub at /ai-tools/ or the <a href="/blog/">blog</a> for longer workflows.
Processes data instantly with no server round-trips.
Your data never leaves your browser. Nothing is uploaded.
Works in any modern browser. Nothing to download or install.
No limits, no sign-up, no credit card required.
Works on desktop, tablet and mobile devices.
Beautiful in both themes. Your preference is saved.
Keyboard shortcut
Use Prompt Eval Rubric to draft, estimate, or evaluate prompts before you paste them into a model chat.
Share prompt templates via notes or Recipes without putting API keys in the browser.
Local helpers do not upload prompts to OneDevToolkit for processing.
Answers for this tool. For site-wide help, open the FAQ hub.
Many AI helpers on OneDevToolkit are local prompt builders, estimators, or rubrics. They do not call paid model APIs unless a tool explicitly says so.
Local tools do not upload prompts for training. If you later paste outputs into a third-party chat model, that vendor’s policy applies.
Token and cost estimators are approximations. Verify against your provider’s pricing calculator before budgeting.
Copy prompts into your notes, Vault, or Recipes. Clear the page when finished with sensitive instructions.
Be specific about role, constraints, and output format. Iterate with smaller samples before long documents.
Share final prompt templates via your internal wiki. Avoid putting API keys into browser tools.
Prompt craft tools work offline after load. Live model calls (if any) need network access.
You are responsible for how generated content is used. Follow your organisation’s AI policy.
Suggest improvements through Contact - include the tool URL and a redacted example.