Plan how a known token limit could be divided among prompt sections. Enter a context limit and your own estimates for system instructions, conversation history, tool definitions, and the current request. The planner runs in this browser and does not count or upload text.
Plan a prompt budget
Use the limit and token estimates from the model or application you are working with. Values below are only your planning inputs; this page does not supply model defaults.
Enter a context limit to see the allocation.
- Input allocation total
- 0
- Input space after output reserve
- —
- Unallocated input space
- —
How should I use the result?
Enter counts from the model's tokenizer or your application's usage report when available. If you estimate a section, label it as an estimate in your notes. The output reserve is subtracted from the total context limit before the planner reports space for input. The remaining amount is arithmetic only; it is not a guarantee that a request will fit or run.
Context accounting can include more than the visible prompt text. OpenAI's tokenizer explains tokenization as model-specific text units and gives a rough English rule of thumb; Anthropic documents context awareness as tracking a model's remaining token budget. Use the provider or application that will run your request as the authority for actual counts and limits.
What does the planner calculate?
It sums the four input allocations, subtracts the output reserve from the context limit, and compares the input total with that remaining capacity. It shows remaining space when the plan is under capacity and an overage when it is over. It does not tokenize text, include hidden system or tool overhead automatically, estimate cost, or contact a model API.
Frequently asked questions
Does this tool count my prompt text?
No. Enter token counts you already have. This tool has no tokenizer and does not read or upload prompt text.
Does the example describe a real model limit?
No. The example is an arbitrary arithmetic exercise. Replace every number with values appropriate to your model and request.
Can I use the remainder as a guaranteed context window?
No. Providers and applications may count message formatting, tool schemas, cached input, multimodal content, or other overhead differently. Check the documentation and usage reporting for the actual model and API you use.
What evidence supports this tool?
A public competitor report observed estimated traffic for GPT for Work and keyword traffic for “tokenizer” and “token counter.” These are third-party estimates for a related token workflow, not search volume for this planner, first-party analytics, or proof that this page will receive traffic.
Sources · checked November 3, 2026
Related: coding-agent task brief builder, candidate evidence card builder, and AI coding tool price sources.
Methodology and limitations
The planner sums four user-supplied input allocations, subtracts a user-supplied output reserve from a context limit, and displays remaining space or overage. Its example values are arbitrary. It does not tokenize text, fetch model limits, estimate prices, or guarantee that a request will fit. Competitor and keyword traffic figures cited as demand evidence are third-party estimates for a related token workflow, not search volume or first-party analytics.
Affiliate placeholder {{AFF_TOOLNAME}} — referral link not configured.