HolaClaw
Show HN: In-browser real LLM token counter and cost estimation
Last verified:
What is HolaClaw?
Token Studio is a free token counter and cost estimation tool from HolaClaw that lets you paste any text to see how it tokenizes across different AI model families. The tool provides live token counts, cost estimates, and context-window usage analysis for major LLM providers including Claude (Opus, Sonnet, Haiku), GPT-5 and other OpenAI models, Gemini Pro, and more.
Key features include real-time tokenization as you type, support for multiple model comparision side-by-side, character/word/sentence counting, context window tracking, and estimated API cost breakdowns for both input and output tokens. Everything runs entirely locally in your browser with no data uploaded to any server, ensuring complete privacy for your prompts and documents.
Token Studio is designed for AI developers, prompt engineers, content creators, and anyone working with LLMs who needs to estimate token usage and API costs before making actual API calls. It helps users optimize prompts for cost efficiency, understand how different tokenizers work across models, and plan their AI budget by seeing exact pricing per million tokens for each provider.
HolaClaw pricing
Pricing model: Freemium
Token Studio is completely free to use with no paid tiers. The tool requires no account signup, no subscription, and no API key to use the token counter features. It runs entirely in your browser with a one-time 10.9 MB tokenizer download. While the tool itself is free, it displays estimated API costs based on current pricing from cloud providers like Anthropic, OpenAI, and Google - but you only pay those costs when making actual API calls to those providers, not to HolaClaw.
HolaClaw pros
- Runs entirely locally in browser with zero data uploaded
- Free to use with no account required
- Supports Claude Opus/Sonnet/Haiku tokenizers
- Supports GPT-5 and OpenAI model tokenizers
- Supports Gemini Pro tokenizer
- Real-time tokenization updates as you type
- Side-by-side model comparison for same text
- Live cost estimates for input tokens
- Live cost estimates for output tokens
- Context window usage tracking
- Character count with and without spaces
- Word count display
- Sentence count display
- Shows tokens per word metric
- Shows characters per token metric
- 10.9 MB one-time tokenizer download size
- No terminal or setup required
HolaClaw cons
- Only supports macOS (Windows/Linux coming soon)
- Requires Apple Silicon M1 or newer
- Requires macOS 14 Sonoma or later
- Tokenizer files must download (10.9 MB)
- Limited to web browser usage only
- No mobile app version available
- Cannot batch process multiple texts
- No API integration for automation
- Output cost is only an estimate not actual
- No saved history of past analyses
Frequently asked questions about HolaClaw
Is my text uploaded to any server when using Token Studio?
No, everything runs entirely locally in your browser. Nothing is uploaded to any server, ensuring complete privacy for your prompts and documents.
Which AI models does Token Studio support for tokenization?
Token Studio supports Claude Opus (claude-v4), GPT-5, Gemini Pro, and will add more models including Claude Sonnet, Haiku, and other OpenAI and Google models.
Do I need to create an account to use Token Studio?
No account is required. Token Studio is free to use with no signup forms, no subscription, and no API key needed for the token counter features.
How accurate are the cost estimates shown in Token Studio?
The cost estimates use current pricing per million tokens from cloud providers like Anthropic, OpenAI, and Google. Input and output tokens are priced separately, with output tokens typically costing 4-5x more than input tokens.
What metrics does Token Studio display besides token count?
Token Studio displays characters (with and without spaces), word count, sentence count, tokens per word, characters per token, context window usage, and estimated input/output costs.
Can I compare multiple models with the same text?
Yes, Token Studio allows you to compare models side-by-side using the same text. The interface shows 'Compare models — same text' and displays token counts and costs for each model in a table.
Does Token Studio work on Windows or Linux?
Currently Token Studio via HolaClaw is only available for macOS (version 14 Sonoma or later, Apple Silicon M1 or newer). Windows and Linux support is coming soon.
How do I use Token Studio to estimate my API costs?
Paste or type your text into the input area, select the AI model you want to estimate for from the available tokenizers, and view your results including token count, character/word count, and estimated API cost displayed instantly.
What is the file size for the tokenizer download?
The tokenizer download is 10.9 MB and is a one-time download that loads when you use Token Studio for the first time with each model family.
Can Token Studio help me optimize my prompts for cost?
Yes, by showing real-time token counts and cost estimates as you type, Token Studio helps you understand how different prompt variations affect token usage and API costs, enabling you to optimize for cost efficiency before making actual API calls.