Pay less. Get more.

Reduce your LLM token costs by 54%. Works with Claude Code and Claude Desktop.

0
Tokens saved/month
$0
Cost saved/month
1:6.7
ROI factor
Get Started Free How it works

The problem every AI power user knows

Token costs are rising

Every session consumes thousands of tokens for context that's rarely needed.

Compacting kills quality

When context gets compressed, important details are lost.

Manual optimization doesn't scale

Manually trimming CLAUDE.md? For every project? That doesn't scale.

3 steps to lower token costs

1

Install the plugin

One command. Works with Claude Code (hooks) and Claude Desktop (MCP).

npm install -g @simosphere/tokenizer-speedup
2

Rules match your context

YAML-based rules load only the context your prompts need.

3

See your savings

Real-time dashboard shows token savings and ROI.

Integration into your tooling — step by step.

Claude Code

Fully supported

Automatic hook integration. The plugin injects optimized context with every prompt.

npm install -g @simosphere/tokenizer-speedup
tsu setup --api-key YOUR_KEY
Restart Claude Code — done.

What happens: With every prompt, the plugin loads only the relevant context. You see savings in the dashboard.

Claude Desktop

Available (v0.8.0)

MCP server for Claude Desktop. Optimizes context automatically with every prompt.

npm install -g @simosphere/tokenizer-speedup
{"mcpServers":{"tokenizer-speedup":{"command":"tsu-mcp"}}}
Restart Claude Desktop — done.

What happens: Claude Desktop can now optimize context, analyze tokens and show savings via MCP tools.

ChatGPT / OpenAI

Token analysis available

Local token analysis for OpenAI models. No live integration into ChatGPT sessions.

tsu analyze src/app.ts --provider openai --model gpt-4o

Counts tokens locally with the OpenAI tokenizer. Ideal for cost estimates before sending.

Configuration

Create a project-local tokenizer-speedup.yaml for custom rules.

tsu init

Defines keyword triggers, file patterns, and token limits. Each project can have its own rules.

Your data works for you — in every scenario.

Daily Feature Development

You use Claude Code or Claude Desktop daily for new features. TokenizerSpeedUp loads only the files your prompt needs — not the entire repository.

Before 180K Tokens
After 108K Tokens
-40%
Free tier is enough →

Code Reviews in a Team

5 developers review PRs and refactor code with Claude Code and Claude Desktop. Context Gating via hooks and MCP prevents old branches and irrelevant modules from being loaded.

Before 750K Tokens/mo
After 345K Tokens/mo
-54% $36/mo saved
Pro from $5/mo →

Multi-Project Token Management

15 developers work across multiple client projects. Session Intelligence compresses context across projects and delivers budget reports per team.

Before 3M Tokens/mo
After 1.2M Tokens/mo
-60% $180/mo saved
Team from $15/seat →

Simple pricing. Pay only when you save.

Free

Ideal for solo devs
$0
/month
  • ✓ Up to 100K tokens saved/month
  • ✓ Basic dashboard
  • ✓ Default YAML rules
  • ✓ Local metrics
Get Started Free

Pro

Popular
For growing teams
$5
/month
  • ✓ Unlimited token optimization
  • ✓ Full analytics dashboard
  • ✓ Custom YAML rules
  • ✓ Priority support

10% of savings above 500K tokens

Start Pro

Team

Enterprise-ready
$15
/seat/month
  • ✓ Everything in Pro
  • ✓ Team-wide rule sync
  • ✓ Per-seat billing
  • ✓ Admin dashboard

8% of savings above 1M tokens

Contact Sales

54% token reduction in 2 weeks

SIMO GmbH

54% token reduction in 2 weeks

Before
550 Zeilen / ~5.000 Token
After
251 Zeilen / ~2.300 Token
Monthly savings
$36
ROI
1:6.7

Privacy first. Your prompts never leave your machine.

Local

The plugin runs entirely on your machine.

Numbers only

Only token counters are transmitted. No prompts, no code.

Transparent

Only token counters are sent. You stay in full control.

FAQ

What is TokenizerSpeedUp?
A plugin for Claude Code and Claude Desktop that automatically optimizes your token usage through intelligent context management.
How much can I save?
Typically 30-60% of token costs. Our own test showed a 54% reduction.
Is it safe?
Yes. The plugin runs locally on your machine. Only anonymous token counters are sent to our dashboard.
Does it work with other LLMs?
Works with Claude Code (hooks) and Claude Desktop (MCP). Local token analysis also supports OpenAI models.
Can I cancel anytime?
Yes, monthly cancellation with no minimum term.
How do I install the plugin?
Two commands: npm install -g @simosphere/tokenizer-speedup, then tsu setup --api-key YOUR_KEY. Done.
Does it work with Claude Desktop?
Yes! Since v0.8.0, TokenizerSpeedUp includes an MCP server. Add it to your claude_desktop_config.json and Claude Desktop can optimize context automatically.
Does it work with ChatGPT?
Partially. Token analysis works locally: tsu analyze --provider openai. Live integration into ChatGPT sessions is not possible.
What is the difference between tokens and estimate?
tokens = exact BPE tokenizer count (same as the LLM API). estimate = fast word heuristic (words × 1.4). For billing purposes, always use tokens.
What is tokenizer-speedup.yaml?
An optional configuration file in your project root. Defines which files are loaded for which keywords. Create with: tsu init
What MCP tools are available?
Three tools: optimize-context (context gating for prompts), analyze-tokens (token counting for text and files), and get-savings (cumulative session savings).
How do I configure the MCP server?
Install globally via npm, then add {"mcpServers":{"tokenizer-speedup":{"command":"tsu-mcp"}}} to your claude_desktop_config.json. Restart Claude Desktop.

Your Dashboard

Real-time insights into your token savings

Coming soon
Dashboard preview with sample data
12.4M
Tokens saved
$36
Cost saved
247
Sessions
54%
Savings
Daily Savings (Last 7 Days)
Mo
Di
Mi
Do
Fr
Sa
So
Tokens in millions 2.6M peak