Overview
Proxy analytics dashboard
Snapshot
Tokens Saved
0
original → optimized
Output Tokens (LiteLLM)
0
0 prompt · 0 completion
Avg Savings / Request
$0.0000
per proxied request
Requests Processed
0
routed via smart model selection
💸 What You'd Pay Each Provider (vs Tokenly)
🎯 Complexity Distribution
📈 Total Spend — Tokenly vs. Standard Providers
🗜️ Tokens Saved Per Day
⚡ Recent Requests
Request Log
Full history of routed AI requests with token and cost breakdown
All Requests
| Time ↕ | Prompt | Model | Provider | Tier | Tkns | Actual | GPT-5 | Sonnet | Gemini | Grok | Saved | Lat. |
|---|
Benchmark Analysis
Deep dive into cost efficiency and performance gains
Using Tokenly Proxy
Actual Spend
$0.00
Tokens Billed
0
Avg. Latency
0ms
Without Tokenly (Frontier Only)
Hypothetical Spend
$0.00
Raw Tokens
0
Frontier Models
GPT-5 / Opus
📊 Per-Provider Savings Breakdown (vs your usage)
Connect Your IDE
Configure your editor to use Tokenly as the AI provider
Continue.dev
Tokenly automatically updates your Continue config on start. Just select "Tokenly" from the model dropdown.
{
"title": "Tokenly",
"model": "tokenly",
"apiBase": "http://localhost:8001/v1",
"provider": "openai"
}
Cursor / VS Code
Go to Settings -> Models -> OpenAI and override the base URL.
Base URL: http://localhost:8001/v1
API Key: tokenly
Any OpenAI SDK
Point your custom tools or scripts to the Tokenly proxy port.
client = OpenAI(
base_url="http://localhost:8001/v1",
api_key="tokenly"
)