LLM Cost Estimator
Estimate multi-model API cost from tokens or pasted prompts.
OpenFill in a few fields and get a prompt with role, task, context, tone, constraints and format.
Type a persona and pick one of five tones.
Add the task and any context the model needs.
Say how the answer should look, such as a table or JSON.
Click Generate, then copy the sectioned prompt.
You keep typing one-line requests into a chatbot and getting vague, generic answers back. The fix is usually not a cleverer sentence but a prompt with clear sections. Prompt Generator turns a role, a task and a target format into that structure.
Enter a role or persona, choose a tone (professional, friendly, concise, technical or creative), describe the task, optionally add context, and say what format you want back. Click Generate and the output is assembled as Markdown with Role, Task, Context, Tone, Constraints and Output format headings. The constraints section always includes three baseline rules: be accurate and actionable, state assumptions when information is missing, and do not invent facts. Leave the role blank and it falls back to 'an expert assistant'; leave the format blank and it asks for a clear structured response with headings.
The context field is where most prompts improve. Mention the stack, the audience, or what you have already tried, since that is information the model cannot guess. If you reuse the same shape with different inputs, move it into the Prompt Template Builder and swap values with placeholders. For ready-made starting points by task, browse the prompt library. To check an existing prompt instead, the Prompt Optimizer scores it against the same ideas.
The prompt is built from your fields inside the page with simple text assembly. No AI model is called to write it.
Keyboard shortcut
Answers for this tool. For site-wide help, open the FAQ hub.
No. It places your inputs into a fixed template with labeled sections. That keeps the output predictable, and you can edit it freely after copying.
The three default constraints are always included in the output. Copy the prompt and edit or extend that section in your chat tool, or put extra rules in the Context field so they appear in the generated text.
Any chat model that accepts plain text. The Markdown headings are readable by ChatGPT, Claude, Gemini and local models alike, and they also make the prompt easier for teammates to review.