Unlimited tokens. Literally.

Coding agent that runs in your terminal on dedicated GPUs. Fixed price per dev, unlimited tokens. Start using it now.

147.8Btokens served this month

terminal

Install in your terminal

$ npm i -g @verboo/code

A tool that cuts off your train of thought mid-code isn't a tool. It's a roadblock.

Token caps are the Achilles' heel of every "AI coding tool" out there. You build context, the model starts understanding your codebase, and right when it's about to solve the real problem. BAM. Limit hit. Wait, reload, pay more. Or switch tools.

The whole category put a clock on the developer's face. We think one clock on your wrist is enough.

// recent inferences

—aggregate infrastructure tok/s

awaiting connection...

The shortest path between your code and a GPU that shares resources with no one.

Your terminal or IDE → OpenAI-compatible endpoint → our dedicated GPU. Straight path, no middlemen. Your code never touches OpenAI or Anthropic. Runs on Brazilian servers, on a GPU allocated to your plan.

The models are open source. You know what runs, where it runs, and how it runs. Transparency is a value, not a slogan.

your terminal / IDE

OpenAI-compatible endpoint

our dedicated GPU

13 models. Each with its own style.

From the flash that responds in milliseconds to the war tank with 1M context. Switch between them mid-conversation without losing the thread. All open source, all running on our GPUs.

deepseek-v4-flash

1M ctxDestaque

deepseek-v4-flash-0731

1M ctxDestaque

deepseek-v4-pro

1M ctx

deepseek-v4-pro-0813

1M ctx

glm-4.7-flash

201k ctx

glm-5.2

1M ctx

glm-5.3-flash

1M ctx45 tok/s

kimi-k2.7

1M ctx

mimo-v2.5

1M ctx60 tok/s

mimo-v2.5-pro

1M ctx36 tok/s

minimax-m3

1M ctx85 tok/s

qwen3.6-27b

262k ctx

qwen3.8-27b

262k ctx

Switch with /model mid-conversation. No context lost.

Pick your plan. No surprises at the end of the month.

Fixed pricing in your regional currency, with Pix or card where available. All plans include truly unlimited tokens. Rate limits vary because heavier models demand more processing. We'd rather be honest about it.

13 models total on the platform. Each plan unlocks a different set.

Loading prices for your region…

Enterprise

100% dedicated GPU cluster, mTLS, contractual privacy

Talk to sales

Usage-based API

Input from $0.14 per 1M tokens. Prepaid balance, no monthly fee, and an OpenAI-compatible endpoint.

View pricing

Refer friends and earn credits.

Every Verboo subscriber can join the referral program. Eligible payments from your referrals generate traceable credits for you.

Subscribe to Verboo

After subscribing, you access your referral link inside the platform.

Invite other devs

Share your link with friends, community, clients or professionals who use AI daily.

Accumulate credits

Each eligible payment generates credits at the current rate, with a configurable safety holding period.

Request a Pix payout

Once credits are available, request your balance and track the manual payout in your dashboard.

Every credit keeps its payment source, rate, and release date for a transparent history.

Subscribe and activate my link

An active community using AI daily.

Join a community of devs, founders and creators who are already using Verboo to code, automate, test ideas and accelerate real projects. Over 100 devs in the group, space to ask questions, share experiences, keep up with news and see real use cases.

Partnerships with tech creators: shoutouts, coupons and exclusive content.

Join the Discord Talk about partnership

Dev orchestrating a squad of agents with Verboo Code and celebrating in the group

Member saying they never hit rate limit even with 3 terminals open

The community declaring their love for Verboo Code

New address. New API key. Same client.

OpenAI-compatible endpoint. Swap the URL and API key. Done. Keep using Cursor, VS Code, JetBrains, OpenClaude or any client of your choice. The difference is now the token is infinite.

curl https://code.verboo.ai/router/v1/chat/completions \
  -H "Authorization: Bearer $VERBOO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<MODEL_ID>",
    "messages": [{"role": "user", "content": "Refactor this component to hooks."}]
  }'

Read the complete documentation

Questions we've already answered. But we'll answer again.

Are unlimited tokens real or is there fine print?

Real. No token cap on any plan. What exists is rate limit per minute, because heavier models demand more GPU. We'd rather be honest about it than sell "unlimited" and choke mid-stream. Individual use, no reselling. To integrate into a product or pipeline, use the prepaid token API.

Do you train on my code?

No. Your code never leaves our server. Requests go from your machine straight to your plan's dedicated GPU. Never touches OpenAI or Anthropic. We don't train models on your code or review your calls. Logs and audit are yours.

Is the CLI open source?

Yes, the CLI is a fork of OpenClaude, available on GitHub (@verboo/code). What runs on the server (models, endpoint) isn't open source; but the models that run are.

Price comparison with Cursor or Copilot?

Cursor and Copilot have their own usage and billing policies. With Verboo Code, every plan has a fixed monthly price shown in your regional catalog, with no additional per-token charge. Speed limits and included models are clear before you subscribe.

Does it work with VS Code, Cursor, JetBrains?

Yes. The endpoint is OpenAI-compatible. You configure the URL (code.verboo.ai/router/v1) and the API key in whatever client you already use. Done.

What's the difference between subscriptions, usage-based API, and Enterprise?

Junior / Pro / Max / Ultra: for individual use in the CLI, IDE, or an OpenAI-compatible client. Unlimited tokens with rate limits based on the plan.

Usage-based API: prepaid balance for product and pipeline integrations. Each model charges input, cached input, and output separately, with no monthly fee.

Enterprise: 100% dedicated GPU cluster, mTLS, no rate limits, and contractual privacy for teams that need total isolation.

Why would I pay for Verboo Code instead of ChatGPT?

Because they're different products. ChatGPT is a web assistant: it doesn't run in your terminal, doesn't access your repository, doesn't hold long sessions of hundreds of thousands of tokens. Verboo Code is a coding agent with unlimited tokens working inside your code. If you just need quick answers, free ChatGPT works. If you code every day, Pro costs less than a day of a mid-level dev's work.

Is there a catch with unlimited tokens?

No. The price is fixed, the limits are clear, and you're not charged per token at the end of the month. What exists is rate limit per minute, which varies according to the model's weight. It's an engineering matter, not hidden cost. We'd rather explain how it works than sell "unlimited" without transparency.

Can I resell or share my access?

No. Unlimited tokens are for individual use. Each subscription covers one developer. Rate limits prevent sharing and abuse. For resale or product integrations, use the prepaid token API; for exclusive capacity, choose an Enterprise cluster.

$ npm i -g @verboo/code

Get started now. No hassle.

Runs in your terminal in 30 seconds. If you don't like it, just stop using it. No bureaucracy, no end-of-month surprises.

Paste in your terminal and Boo installs everything. 30 seconds.

Prefer the desktop app? Download