free tool · pricing table 2026-10-26 · works without JavaScript

LLM tokenizer and compressor

Count tokens for five providers and preview the lossless whitespace compaction the gateway applies to tool schemas.

The sample two-tool schema is 262 tokens on the OpenAI tokenizer, 382 on Anthropic's and 276 on Gemini's; the gateway's lossless compaction (whitespace runs inside descriptions only) brings the OpenAI count to 143, 119 tokens (45%) saved on every call that carries it. Names, enums, defaults and examples are never touched.

As of 2026-10-26 · Source: pricing table 2026-10-26 and the gateway's own tokenizer and benchmark files; your inputs never leave the browser except token counts, which are not stored

At a glance

Tokenizers: openai, anthropic, gemini, xai, deepseek
Sample tokens (openai): 262 (computed)
Sample tokens (anthropic): 382 (computed)
Sample tokens (gemini): 276 (computed)
Sample tokens (xai): 276 (computed)
Sample tokens (deepseek): 276 (computed)
After compaction (OpenAI): 143 (computed)
Saved per call: 119 tokens (45%) (computed)
What compaction changes: whitespace runs inside description strings
What it never changes: names, enums, defaults, examples, patterns, required lists

Sample tool schema: tokens per provider tokenizer, before and after lossless compaction

Tokenizer	Before	After	Saved
openai	262	143	119
anthropic	382	209	173
gemini	276	151	125
xai	276	151	125
deepseek	276	151	125

Tokenizer and compressor

Text, prompt or tool-schema JSON

[
  {
    "type": "function",
    "function": {
      "name": "lookup_order",
      "description": "Look   up an order by   its id.   Returns  status, items and the   shipping ETA.  ",
      "parameters": {
        "type": "object",
        "properties": {
          "order_id": {
            "type": "string",
            "description": "The   order id,   e.g.  ORD-12345   "
          }
        },
        "required": [
          "order_id"
        ]
      }
    }
  },
  {
    "type": "function",
    "function": {
      "name": "issue_refund",
      "description": "Issue a   refund   for an order.    Needs an   approved   ticket.",
      "parameters": {
        "type": "object",
        "properties": {
          "order_id": {
            "type": "string"
          },
          "amount_usd": {
            "type": "number",
            "description": "Amount in USD,   at most the   order total"
          }
        },
        "required": [
          "order_id",
          "amount_usd"
        ]
      }
    }
  }
]

Questions people ask

Is the compaction lossless?

Yes. Only runs of whitespace inside description strings are collapsed to one space; the words, their order, and every other field are unchanged, and the cache-tier guard skips compaction when it would drop a prefix below the model's cache minimum.

Why do the five counts differ?

Each provider tokenizes differently; the gateway uses a per-provider factor calibrated on the official tokenizers so audits, receipts and this tool agree with the provider's usage numbers.

Is my text stored?

No. It is counted, optionally compacted, returned and discarded.

What does TokenZero cost?

Free up to $500 per month of routed spend; then 15% of verified savings with a 3x-ROI-or-$0 rule. Gateway overhead is 1.52 ms p50, measured.

Tool-schema bloat playbook Prompt compression with the OpenAI SDK LLM cost calculator Prompt caching savings estimator Cascade routing simulator GitHub repo token-waste auditor Gateway latency and TTFB benchmark OpenAI to Claude cost converter

All tools

The gateway compacts every tool schema on every call: swap base_url and read X-TokenZero-Compaction-Tokens-Saved.Start free