Paste text below, pick a model, and get an estimated token count and approximate API cost. Token counts are heuristic with a low/high range — exact tokens depend on the provider's tokenizer. Nothing is uploaded.
| Provider | Model | Input / 1M | Output / 1M | Context | Verified |
|---|
Prices shown are standard-tier list prices in USD per million tokens, taken from the provider's public pricing pages. Batch, cached-input, and Bedrock/Azure markup tiers are excluded. Embedding, fine-tune, and audio/image prices excluded.
Each model family uses a different tokenizer. As a back-of-envelope:
o200k tokenizer averages ~3.8 characters per token for English prose.The displayed range applies a ±10% margin on top of the model-specific chars-per-token average to give a defensible band.
Is the token count exact?
No — different model families tokenize the same text differently. For an exact count use the provider's
official tokenizer (OpenAI's tiktoken, Anthropic's count_tokens endpoint, or
Google's Gemini count-tokens API).
How is the cost calculated?
Estimated input tokens × the model's per-million input rate, plus the expected output tokens × the per-million output rate. Cached-input and batch-API discounts (often 50%) are not applied.
Are these prices current?
The table shows the date each row was last verified. LLM prices change often — always cross-check with the provider's pricing page before relying on the number for forecasting.
Is my text uploaded?
No. Token estimation and cost math run entirely in your browser. No third party — including the LLM providers themselves — receives the text.