Analyze Claude token usage.
Measure prompts, documents, and images against current Claude models without uploading your content.
Text, code, PDF, DOCX, images, audio, or video · 12 MB per file
Your input, measured across providers.
Choose a provider card to inspect every available model.
Workload & Scaling Planner +
A measurement, not a guess disguised as fact.
Claude text uses the deterministic Provider-Calibrated UTF-8 Token Projection because Anthropic does not distribute its production tokenizer as a lightweight offline browser package. Images use Anthropic’s published 28×28 visual-patch rules and applicable model cap.
Every result identifies its method and accuracy level. Pricing is refreshed from Models.dev with the bundled registry available as an offline fallback.
Read the complete methodology →How this Anthropic token calculator works
Claude text uses the deterministic Provider-Calibrated UTF-8 Token Projection because Anthropic does not distribute its production tokenizer as a lightweight offline browser package. UTF-8 byte length is normalized by an Anthropic-family bytes-per-token coefficient.
Punctuation, code, JSON, Unicode composition, non-English text, and API message framing can shift the final Claude usage reported by Anthropic.
Claude image tokens use Anthropic’s published 28×28 visual-patch method and the applicable model cap. Only dimensions are read locally.
Read the official Anthropic documentation ↗Models included in this counter
Registry reviewed 2026-10-02
| Model | Tokenization method | Context window | Input / 1M | Output / 1M |
|---|---|---|---|---|
| Claude Sonnet 5.5 | Provider-Calibrated UTF-8 Projection | 1,000,000 | $2.00 | $10.00 |
| Claude Opus 5.5 | Provider-Calibrated UTF-8 Projection | 1,000,000 | $4.00 | $20.00 |
| Claude Fable 5.1 | Provider-Calibrated UTF-8 Projection | 1,000,000 | $10.00 | $50.00 |
| Claude Opus 5 | Provider-Calibrated UTF-8 Projection | 1,000,000 | $5.00 | $25.00 |
| Claude Sonnet 5 | Provider-Calibrated UTF-8 Projection | 1,000,000 | $2.00 | $10.00 |
| Claude Opus 4.8 | Provider-Calibrated UTF-8 Projection | 1,000,000 | $5.00 | $25.00 |
| Claude Haiku 4.5 | Provider-Calibrated UTF-8 Projection | 200,000 | $1.00 | $5.00 |
Rates shown are USD per 1,000,000 tokens, verified against the public pricing catalogue on 2026-10-02. The table refreshes from the same live source the calculator uses. Provider invoices and official documentation remain authoritative.
Measure against the model you use
Compare cost with the same real workload
Compare approaches, not just models
Frequently asked questions
How does the Claude token projection work?+
The engine computes UTF-8 byte length and applies an Anthropic-family bytes-per-token coefficient through a deterministic, reproducible projection. Anthropic’s production tokenizer is not distributed as a lightweight offline browser library.
How are Claude image tokens calculated?+
Image dimensions are processed locally with Anthropic’s published 28×28 visual-patch method and the applicable model cap.
Why can Claude API usage differ from this projection?+
API message framing, tool schemas, system instructions, punctuation, code, Unicode composition, and non-English content can change the final token usage reported by Anthropic.
Does the calculator upload content to Anthropic?+
No. The privacy-first calculation pipeline executes locally and does not call the Claude API. Anthropic receives nothing from this page.
What is a token in an AI model?+
A token is a small unit of content processed by a language model. It may be a whole word, part of a word, punctuation, whitespace, or—in multimodal models—a unit assigned to an image. Different providers can split the same input differently.
How is LLM API input cost calculated?+
Input cost is calculated as input tokens divided by one million, multiplied by the selected model’s input price per million tokens. Provider tiers, cached input, tools, and message wrappers can affect final billing.
Does this tool upload my prompt, document, or image?+
No. Text, PDF and DOCX extraction, image-dimension reading, and token calculations run in your browser. Zero prompts, document contents, or uploaded images are sent to TokenCalculator.dev or another calculation server.
Are the token counts exact?+
OpenAI text counts use the matching local BPE encoding. Other providers are clearly labeled estimates unless they publish a deterministic local formula. Image counts use provider-published rules.
How do I count tokens in a PDF or Word document?+
Attach a text-based PDF or DOCX file and the browser extracts its readable text locally, then counts it with every selected provider model. Scanned or encrypted PDFs require OCR and are not supported.
Can I calculate image tokens and vision-model cost?+
Yes. Add a PNG, JPEG, WebP, or GIF. For supported vision models, the calculator reads its pixel dimensions locally and applies the provider-published image token formula. It reports unavailable instead of inventing a value when a provider has no documented rule.
Why can the API report a slightly different count?+
Providers can add message wrappers, system instructions, tools, or model-specific processing. The accuracy badge and methodology explain what this calculator includes.