Text Token Count & Cost Estimate (by LLM Model)

{
  "type": "object",
  "required": [
    "text"
  ],
  "properties": {
    "text": {
      "type": "string",
      "description": "The text to estimate a token count and cost for."
    },
    "model": {
      "type": "string",
      "description": "Model to estimate against, e.g. 'gpt-4o', 'claude-3-haiku' (optional). An unrecognized model name is not an error -- see outputSchema's `model_recognized`."
    }
  },
  "additionalProperties": false
}

{
  "type": "object",
  "properties": {
    "model": {
      "type": "string",
      "description": "The model name the estimate was computed for, echoed back as submitted."
    },
    "text_length": {
      "type": "number",
      "description": "Length, in characters, of the submitted text."
    },
    "token_count": {
      "type": "number",
      "description": "Estimated token count for the submitted text, for the given model."
    },
    "model_recognized": {
      "type": "boolean",
      "description": "Whether the submitted `model` matched one of the estimator's known model profiles. Observed `false` on an unrecognized model name -- that call still returns a normal estimate (using a fallback chars-per-token profile), not an error."
    },
    "cost_per_1k_tokens": {
      "type": "number",
      "description": "The per-1,000-token input rate used to derive `estimated_cost_usd` for this model."
    },
    "estimated_cost_usd": {
      "type": "number",
      "description": "Estimated dollar cost of sending this text as input to the given model, derived from `cost_per_1k_tokens`."
    },
    "fits_context_window": {
      "type": "boolean",
      "description": "Whether `token_count` fits within `max_context_for_model`."
    },
    "chars_per_token_used": {
      "type": "number",
      "description": "The characters-per-token ratio used to derive `token_count` for this model. Observed 4 for a recognized model (gpt-4o) and 3.8 for an unrecognized one -- varies by model and by recognition, not a fixed constant."
    },
    "max_context_for_model": {
      "type": "number",
      "description": "The model's maximum context window size, in tokens."
    }
  },
  "description": "The token-count estimate, as returned at routing.responseResultPath ($.output).",
  "additionalProperties": true
}