Tokelix
Tokelix / TOKEN ROUTERFor developers & teams

Your AI calls.
Your control.

From your first line of code to your whole team. Connect to LLMs through one API, with routing, usage and costs in one workspace.

Start buildingExplore the docs
YOUR API ENDPOINT
BASE URL
https://router2.mvu.app/v1
Usage-based billingTeam workspaceBudget controls
Tokelix WORKSPACEREQUEST FLOW · ILLUSTRATION
Your apps & AI toolsCodex · Claude Code · SDK
01
API REQUEST
TokelixOne endpoint. One workspace.
02
RouteBudgetMeter
MODEL ROUTING
GPT
Claude
DeepSeek

Build your product. Keep every token accounted for.

BUILD WITH CLARITY.Discover models & pricingONE API. YOUR WORKSPACE.

Model pricing · Pay as you go

All prices are USD per 1M tokens. Requests use the route price in effect when they start.

gpt-5-6-lunaOpenAI
Input / 1M tokens$0.042Output / 1M tokens$0.336
Quick start
deepseek-v4-1-flashDeepSeek
Input / 1M tokens$0.12Output / 1M tokens$0.474996
Quick start
gemini-3-8-flash-openaiGoogle
Input / 1M tokens$0.225Output / 1M tokens$1.125
Quick start
gemini-3-7-flash-openaiGoogle
Input / 1M tokens$0.225Output / 1M tokens$1.125
Quick start
gemini-3-6-flash-openaiGoogle
Input / 1M tokens$0.225Output / 1M tokens$1.125
Quick start
claude-haiku-4-5Anthropic
Input / 1M tokens$0.275Output / 1M tokens$1.375
Quick start
See all model prices
/v1unified API endpoint
3routing strategies
USDtransparent usage billing
UTC+8monthly budget period
01 / CLARITY IN EVERY CHARGE

Every token.
Clearly accounted for.

Input, output and cache are metered separately. From a single request to your team’s bill, every charge has a record.

View balance & billingManual credits · Separate organization wallets
HOW EACH REQUEST IS METEREDUSD
Input tokensWhat you send to the model
Output tokensWhat the model generates
Cached tokensMetered by the route’s cache rules
Bills retain the price at the time of the request

Quick start

Point your existing tools at Tokelix. Standard endpoints, transparent per-request billing.

Full setup guides
Python
from openai import OpenAI

client = OpenAI(
    base_url="https://router2.mvu.app/v1",
    api_key="<TOKELIX_API_KEY>",
)
response = client.chat.completions.create(
    model="gpt-5-6-luna",
    messages=[{"role": "user", "content": "Hello"}],
)
print(response.choices[0].message.content)

Point the base URL at Tokelix, drop in your API key — the rest of your code stays the same.

Running in Production

Spend TrackingUSD

Every request records the model, tokens and actual charge.

Routing Strategies3

Choose routes by price, priority or a custom price cap.

Access & BudgetsUTC+8

Organization and member access, wallets and budgets are enforced.

A Router You Can Build Your Business On

Built for Claude Code, Codex and long agentic sessions.

Model IntegrityEligible routes preserve the requested model and protocol during failover.01
Data PrivacyOrganization ledgers, keys and membership permissions are checked independently.02
Spend TrackingInput, output and cache are metered separately; bills retain historical prices.03

How are routes and prices selected?

One account, one bill. Choose the strategy that fits your team.

Price-first routingEligible routes for the same model are ordered by price.01
Curated routingRoutes follow administrator-configured priorities.02
Custom routingSet provider price caps and use only qualifying routes.03
PRODUCTION ROUTING

Price and availability

Every request checks wallet and budgets. Failover preserves model and price constraints.

Price
Configured route prices
Model
Preserves the requested model
Budget
Organization and member limits
Example routes · not live status
AppTokelix
PRIMARYModel A · Route 1Degraded · not used
FALLBACKModel A · Route 2Serving · selected route
BUILD YOUR NEXT IDEA

Your next idea.
Start with a connection.

One LLM endpoint for Codex, Claude Code and the apps you build.

Read the setup guide
01
Create your API keyManage access in your team workspace
02
Connect your AI toolsCodex · Claude Code · SDK
03
Build, with costs in viewOne place for usage and budgets

Frequently asked questions