GPT-5.4, Claude 4.6, Gemini 3.5, DeepSeek-V4, Qwen3.7 and more side by side.
Every LLM Token Counts.
Real-time LLM prompt token analysis across models. Cut API costs. Stay GDPR & PIPL ready.
Paste a prompt below — see tokens, costs, and savings instantly. 100% local.
Token Analysis
Analyzing tokens…
Why LLM token math matters
Token counting runs in your browser. Zero server-side prompt storage.
Cost gap across models for the very same prompt — almost all of it is price per token, not token count.
Full demo free forever. Paid plans coming — join the waitlist.
How it works
Three steps. No signup. Your prompt stays on your device.
Paste your LLM prompt
Drop in any system prompt, RAG context, or chat template — English, 中文, or mixed.
See token breakdown
Watch real-time token blocks, total count, and Chinese inflation vs OpenAI baseline.
Compare model costs
Instant per-call cost estimates in USD, GBP, or CNY. Pick the most efficient model.
Planned: production-grade LLM infrastructure
On the roadmap — caching, routing, and compliance proxy for teams scaling beyond the free tool.
AI Cache Proxy
Deduplicate identical prompts at the edge. Cut redundant API spend by up to 40%.
Privacy Audit
PIPL, GDPR, and global PII detection before data leaves your region.
Smart Model Routing
Route by cost, latency, or compliance policy across GPT-5.4, Claude 4.6, Gemini 3.5, DeepSeek-V4, and Qwen3.7.
Who uses Tokened?
Real scenarios where token visibility saves money and reduces risk.
Chinese product teams
Chinese text costs about 1.10 tokens per character on GPT-4-era cl100k_base but only ~0.64 on o200k_base, so check which encoding you are on before assuming a tokenizer win — then compare on total cost.
Startups & indie devs
Estimate monthly API spend from your actual prompts before you ship — no surprises on the billing dashboard.
EU / UK / China teams
Audit prompts locally before they hit a US-hosted model. Planned PII pre-check coming in the next release.
Quick Start
Integrate in minutes. All token counting runs locally in the browser.
3-minute integration. Runs locally — your prompts never leave the browser.
npm / TypeScript
npm install gpt-tokenizerimport { encode } from 'gpt-tokenizer'
const tokens = encode(prompt).lengthPython
pip install tiktokenimport tiktoken
enc = tiktoken.get_encoding("cl100k_base")
len(enc.encode(prompt))Guides & insights
Practical LLM token knowledge — more articles coming soon.
5 min read
How Much Do Chinese Prompts Really Cost on OpenAI? (Measured)
Measured: cl100k_base spends ~1.10 tokens per Chinese character, o200k_base only ~0.64. Where the Chinese token penalty still bites, and where the real saving comes from today.
Read article →4 min readHow Prompt Caching Can Cut LLM API Spend by 40%
Identical system prompts resent on every request are wasted spend. Learn how edge caching and Tokened Gateway solve this.
Read article →4 min readGDPR, PIPL, and Why Local Token Analysis Matters
Sending prompts with PII to US-hosted LLMs creates compliance risk. Local-first tooling is the safest first step.
Read article →Roadmap
What is live today, and what is coming next.
Token calculator & cost dashboard
- Real-time token highlighting (OpenAI tokenizers)
- Multi-model cost comparison (GPT-5.4, Claude 4.6, Gemini 3.5, DeepSeek-V4, Qwen3.7)
- Chinese token inflation insights
- USD / GBP / CNY support
- EN / 中文 bilingual UI
Developer plan
Pricing TBD — join the waitlist for early access
- PIPL & GDPR PII pre-check
- One-click prompt slimming
- API access & webhooks
- Priority model price updates
Enterprise & Tokened Gateway
Custom pricing — contact us for enterprise
- Drop-in OpenAI base URL proxy (api.tokened.uk)
- Edge caching & smart model routing
- SSO & team workspaces
- Custom compliance rules & dedicated SLA
Pricing
100% free, no account required. All token analysis runs locally in your browser.
Free
$0forever
- Unlimited local token counting
- Multi-model cost comparison
- EN / 中文 UI
- Chinese token inflation insights
Want paid features early?
Developer and Enterprise plans are on the roadmap — not available to purchase yet. Join the waitlist and we will notify you when they launch.
Get in touch
Interested in beta access, Gateway, or enterprise plans? Email us — no forms or third-party setup required.
hello@tokened.uk
Click the address to open your email client.
FAQ
Common questions about LLM tokens, privacy, and our roadmap.
No. All token counting runs locally in your browser using the gpt-tokenizer library. We do not store, log, or transmit your prompts.
OpenAI counts are exact when using the matching tokenizer (o200k_base for GPT-5.x / GPT-4o, cl100k_base for legacy). Claude 4.6, Gemini 3.5, DeepSeek-V4, and Qwen3.7 costs use estimated token multipliers — always verify against your provider billing.
No. Tokened is exclusively about LLM API tokens — the units OpenAI, Anthropic, and others use to measure and charge for text processing.
Gateway (api.tokened.uk) is in Phase 3 of our roadmap. Join the beta waitlist to get notified when the proxy beta opens.
Not yet. Only the free local analyzer is live. Developer and Enterprise tiers are planned — pricing TBD. Join the waitlist for early access.
Use npm install gpt-tokenizer (TypeScript) or pip install tiktoken (Python). See the Integration section for copy-paste examples.
Privacy Policy
tokened.uk provides browser-local LLM token analysis tools. By default, your prompts are processed entirely in your browser and are not sent to our servers.
If you contact us by email or join our waitlist, we collect only the information you provide (e.g. email address, message content) to respond to your inquiry.
We do not sell personal data. We may use anonymized usage analytics to improve the product.
For GDPR or PIPL-related requests, contact us at the email below.
Last updated: June 2026.
Terms of Service
tokened.uk is provided "as is" for informational and developer tooling purposes. Cost estimates are approximate and based on publicly available model pricing.
Token counts for non-OpenAI models may be estimates. Always verify against your provider's billing dashboard before making production decisions.
You are responsible for ensuring your use of LLM APIs complies with applicable laws, including GDPR and PIPL.
We may update these terms. Continued use of the site constitutes acceptance of the current version.
Last updated: June 2026.