Token Bank Token Bank
Sign in Free download
Personal AI hub · see clearly · spend less · stay simple · get smarter · earn from idle

See clearly. Spend less.
Stay simple.
Get smarter — and earn while idle.

One local hub for Claude / Cursor / Codex and more: onboard, trace, and route in one place—without changing how your tools work. Resources grow from your habits; idle capacity can earn credits.

Free download Baidu Netdisk View on GitHub

macOS · Windows · Linux · Apache-2.0 open source

Gateway Sessions Trace Providers Assets Portrait Playground Usage Circles Contribute Network Tray
localhost:11430
Token Bank Gateway
Gateway · App toolbox
One-click install Claude Code, Kimi Code, Cursor, Codex and more.

Works with the agents you already use

Claude Claude Code
Codex Codex CLI
Cursor Cursor
Kimi Code Kimi Code New
Gemini Gemini CLI
Copilot Copilot
Cherry Studio Cherry Studio
OpenCode OpenCode
OpenClaw OpenClaw
Hermes Hermes
0
One local port
fronts every AI tool
0
protocols auto-converted
Anthropic · OpenAI · Gemini
0×
max quality multiplier
steadier uptime, higher earnings
0 min
auto credit settlement
contribute and earn in real time
Five things · WHAT IT DOES

See clearly · Spend less · Stay simple · Get smarter · Earn from idle

01

See clearly

TRACE · ANALYTICS

Onboard agents; full trace of route, model, tokens, cost. Multi-device inventory with subscriptions vs PAYG side by side.

02

Spend less

SMART ROUTING

Keep native model names; route to local or cheaper supply. Local-first chain, scene strategies, optional lossless compression.

03

Stay simple

ONE-CLICK HUB

One-click onboard/restore; multi-account CLI by directory; tray for status. Point tools at one local address.

04

Get smarter with you

PORTRAIT · FOR YOU

Mine a work portrait from real habits; discover, accumulate, and iterate MCP / Skill / Prompt / Agent that fit you.

05

Earn from idle

COMMUNITY SHARING

Contribute idle capacity to community sharing for credits; circles and a network map; spend credits on shared models.

New in v0.5

From a gateway
to an agent hub.

v0.5 stitches onboarded agents, multi-account CLIs, and community resources into one local control plane, without changing how your tools talk to models.

AGENT ORCHESTRATION

Playground with a main agent

Set a main agent as the aggregation entry. It receives the task, plans steps, and dispatches to other onboarded agents. Tool streams stay visible; stop and resume when you need to.

aggregation entry
→ main agent receives task
→ dispatch to peer agents
→ tools · chunks · stop/resume
MULTI-ACCOUNT CLI

Many logins, one shim

Scan or add Claude / Codex instances. Dispatch by working directory so each project keeps its own account, quota, and route.

GETS SMARTER WITH YOU

Portrait · For You

Mine a work portrait from real habits; discover, accumulate, and iterate MCP / Skill / Prompt / Agent that fit you.

RESOURCE HUB

MCP · Skill · Prompt

Community catalog plus personalized recommendations; project onto runtimes; prompts via tokenbank-prompts MCP.

Connect once

Native model names unchanged,
zero client edits.

In the Gateway tab, installed tools appear automatically. Three clean states: track only · via gateway · reverted.

Swap to any third-party model in a click — no client changes.

01
Track — start counting
Track an app's usage even while it still uses the official subscription
02
Pick a model / scenario route
Config is rewritten automatically; traffic flows transparently through the gateway
03
Revert — one click back
Restore the official config, stop tracking, clean and reversible
Three ways to connect · zero command changes
CLI shim · multi-account
Claude Code · Codex · Kimi Code · Gemini CLI · OpenCode — injects BASE_URL and dispatches by workdir across accounts
Config-file patch
Claude Desktop · Codex Desktop · OpenClaw — point them at the local gateway in one click
OpenAI · Anthropic · Gemini
Text, image & video APIs adapted — Cursor and others via OPENAI_BASE_URL or a dedicated key
$ export OPENAI_BASE_URL=
  http://localhost:11430/v1
Smart routing

One local routing chain, stepping down until it hits.

Each app can bind its own scenario route; the global supply chain is the fallback. When local is unavailable it switches to community sharing — transparent to the agent throughout.

Your AI tools
Claude Code · Cursor · Codex · Kimi Code — native model names unchanged
Token Bank gateway
Transparent rewrite Protocol convert Lossless compress Usage tracking
Local sources
Ollama — zero latency, zero cost Free APIs · Groq / GitHub Models Subscription / pay-as-you-go API
Keys never leave your machine
Community sharing
Shared compute network Online models sync dynamically Spend credits to call remote nodes
Auto fallback when local is down
Policies fallback round-robin weighted latency direct
Observable

A multi-dimensional dashboard,
aggregated across devices.

Desktop, CLI, and server gateways each register as a device; after sign-in, usage is reported and merged in the cloud.

Local / community sharing source split — see owned vs. shared compute at a glance
Compression savings — count, tokens saved, and ratio, merged across devices
Subscription amortized daily + usage estimates, alongside raw token stats
Call logs record every routing result, status, and latency
Supply source split
Last 7 days
0%
local hit
Local sources0%
Community sharing0%
Official subscription0%
By app · tokens & spend
tokens · spend
Claude Code
0 0
Cursor
0 0
Codex CLI
0 0
Kimi Code
0 0
Gemini CLI
0 0
Cherry Studio
0 0
By device · tokens & spend
tokens · spend
MacBook Pro
0 0
Linux Gateway
0 0
Windows PC
0 0
WSL CLI
0 0
Community compute · share & exchange · earn from idle

Personal and community compute, scheduled as one.

Use your own free tiers, subscriptions and metered accounts first; tap community shared compute only when needed — both switch transparently inside one gateway.

01

Personal compute

Supply under your name: local Ollama, vendor free tiers, app/API subscriptions, and metered accounts billed at list price.

Free tier Subscription Pay-as-you-go
02

Community compute

Members contribute idle quota, shared through a credits network. Spend credits to call models others supply — often cheaper than official.

Credit swap Cross-model flow Idle sharing

Cross-model exchange

Contribute quota for models you have, earn credits, and spend them on models you don't.

Time-shifted value

Idle quota becomes credits you bank — draw on them anytime later, across models and hours.

Transparent mutual-aid

Open rules, open source. The contribution rate beats the consumption rate; higher quality earns a higher multiplier.

How earning works

Contribute idle quota,
get spendable credits.

Contribute local Ollama, idle upstream quota, even private models on a corporate intranet — agents dial out over WebSocket, no inbound ports needed.

0.5–1.5×
quality multiplier
~5 min
per settlement
earn > spend
rate design
How credits are calculated
credits = (output_tokens / 1000)
× contribute_rate
× quality(0.5–1.5)
cost = (prompt+completion / 1000)
× consume_rate

Steadier uptime, faster responses, and higher success rates raise your quality multiplier — long-term contributors come out ahead.

Circles · share with peers

Find your people,
share models and credits.

Create or join circles of like-minded AI users. Share contributed models inside the group, invite friends for mutual credit rewards, and contribute to multiple circles at once.

01
Create or join
Start your own circle or join via invite code or link — owner and member roles, member list at a glance.
02
Circle-scoped model sharing
When you contribute compute, pick which circles receive it. Members call shared models through the same gateway — routing stays transparent.
03
Invite peers, earn together
Share an invite link — when a friend joins, both earn credits. Post messages and replies to coordinate with circle mates.
Explore circles →
AI Builders
12 members · you are owner
Invite peers
Members
U A K M +8
Shared models
claude-sonnet-4 ● online
gpt-4o ● online
Messages
Alex
Anyone free to share a Claude route this week?
You
Added mine to this circle — try claude-sonnet-4.
Get started

Spend your tokens where they count.

One local address fronts every AI tool and agent workflow. Double-click to install on macOS / Windows, or run from the CLI on Linux.

Free download Baidu Netdisk View on GitHub
FAQ

Common questions.

How is Token Bank different from calling the API directly? +

Your tools keep their native model names and configs — Token Bank sits in between as one local address. It traces every token, routes to the cheapest viable source, compresses requests losslessly, and converts protocols, all transparently.

Are my API keys uploaded to the cloud? +

No. Keys stay on your machine and are used only by the local gateway. Sign-in syncs usage statistics for cross-device aggregation — never your credentials.

Is community compute safe to use? +

Community routing is opt-in and only used as a fallback. Contributors dial out over WebSocket with no inbound ports exposed; settlement rules are open and credits are auditable, so contribution always out-earns consumption.

Which models and providers are supported? +

Anthropic, OpenAI, and Codex protocols are auto-converted, plus Ollama, Groq, GitHub Models, and any OpenAI-compatible endpoint. You can swap in third-party models per app without touching the client.

Is it free and open source? +

Yes — Token Bank is Apache-2.0 licensed and free to download for macOS, Windows, and Linux / Docker. Community credits are earned by contributing, not bought.

What are Circles? +

Circles are small communities inside Token Bank. Create or join via invite link, share models scoped to your circle, post messages to coordinate, and earn mutual credit rewards when friends join. You can contribute to multiple circles at once from the Contribute settings.

What is agent orchestration in v0.5? +

In Playground, pick a main agent as the aggregation entry. It receives your task, can plan steps, and dispatch work to other onboarded agents. Combined with multi-account CLI and the MCP / Skill / Prompt resource hub, Token Bank becomes a local agent control plane, not only a token proxy.