Short answer: Aider is a free, open source CLI (Apache 2.0) with no model of its own: you bring the key for any compatible provider, OpenAI, Anthropic, Gemini, DeepSeek, or a local model via Ollama, and you pay the token directly to whoever owns the model. Verboo Code is a terminal agent with dedicated GPU, unlimited tokens, and a fixed price, running open models like GLM-5.2 and DeepSeek V4. If you want to pick among the strongest models in the world, already have API credit to spare, or depend on Aider's git-native flow, stay with it. If you have already watched a provider bill spike on a long task, or want to keep Aider as your interface and just swap the engine behind it, that is where Verboo Code comes in, and you can do it without abandoning the tool you already use.
Aider and Verboo Code side by side
| Attribute | Aider | Verboo Code |
|---|---|---|
| Billing model | Free, open source (Apache 2.0), BYOK: you pay per token directly to the provider you choose | Fixed-price subscription, unlimited tokens included in the plan |
| Entry price | Free (the tool itself); the real cost is the provider's token | Junior R$ 99/month |
| Supported models | None built in: any compatible provider (OpenAI, Anthropic, Gemini, DeepSeek, local Ollama) via environment variable | 11 open models (GLM-5.2, DeepSeek V4, Kimi K2.7, and others), switch with /model |
| Editing flow | Git-native: automatic commit on every accepted edit, repo map via tree-sitter | Terminal CLI, no automatic versioning of its own |
| Infrastructure | Runs on your machine, inference happens on the chosen provider's cloud | Dedicated GPU, server in Brazil |
| Context | Varies by model and provider chosen via BYOK | 1M tokens on almost every model (exception: qwen3.8-27b, 262k) |
| OpenAI-compatible endpoint | Native support: OPENAI_API_BASE and OPENAI_API_KEY variables, openai/ prefix on the model name | Exposes its own OpenAI-compatible router, ready for any client to point at |
| Best isolated benchmark | Aider Polyglot, the tool's own benchmark: GPT-5 (high) leads with 88.0% among 60+ models tested (aider.chat/docs/leaderboards, checked 2026-09-09) | GLM-5.2: 62.1 on SWE-bench Pro, ahead of GPT-5.5 (Z.ai, Jun/2026) |
When Aider is the right choice
- You want real git-native behavior: every accepted edit already becomes a commit, with an automatic repo map via tree-sitter.
- You already have API credit with some provider (OpenAI, Anthropic, Gemini) and want the frontier model of the day, without waiting on a third-party integration.
- You want to run 100% local, with Ollama or LM Studio, without depending on any cloud.
- You use architect mode, one reasoning model plans and another edits, and you are not giving that up.
When Verboo Code is the right choice
- You have already watched an API bill spike on a long task and want a fixed price, with no billing surprises.
- You do not want to manage keys and invoices across three or four different providers.
- You want to keep Aider as the interface and just swap the engine behind it, pointing it at Verboo Code's endpoint.
- You want to switch open models mid-session, with
/model, without losing context or swapping keys.
Where each one actually wins
Aider has its own benchmark methodology, Aider Polyglot: 225 Exercism exercises across six languages (C++, Go, Java, JavaScript, Python, Rust), measuring whether the model can apply the right edit format, not just get the answer right. GPT-5 (high) currently tops it, with 88.0% among more than 60 models tested, according to the official leaderboard itself (aider.chat/docs/leaderboards, checked 2026-09-09). None of Verboo Code's 11 models has gone through this specific test yet, so a direct comparison on this exact benchmark does not exist.
On the benchmarks where GLM-5.2, Verboo Code's strongest model, has been measured, the picture is mixed: on Terminal-Bench 2.1 it closes at 81.0 points against Claude Opus 4.8's 85.0, but on SWE-bench Pro it flips, 62.1 against GPT-5.5's 58.6, according to Z.ai's official launch on June 13, 2026. The strongest model on paper changes from benchmark to benchmark, and that is exactly the variation Aider's model-agnostic design lets you route around: switching models is a flag, not a new subscription.
That freedom has a direct cost, though. Claude Opus 5 costs US$ 5 per million input tokens and US$ 25 per million output tokens on Anthropic's official API, according to a September 2026 survey, and every new attempt on a long task adds to that total. Aider itself charges nothing, the real cost lives entirely with the provider. On Verboo Code, the same kind of open model (GLM-5.2, DeepSeek V4) runs with unlimited tokens inside a fixed monthly price, from R$ 99 on Junior to R$ 1,299 on Ultra. The question is not which one thinks better, it is whether the fifth attempt costs more than the first.
And you do not have to pick a side. Aider was built from the start to work with any backend: just export OPENAI_API_BASE and OPENAI_API_KEY and prefix the model name with openai/, exactly as its own official documentation describes. Verboo Code exposes precisely that kind of endpoint, at code.verboo.ai/router/v1. Once configured, Aider keeps auto-committing and building its repo map the way it always has, only the model answering switches to run on Verboo Code, with no token counting.
How to point Aider at Verboo Code's engine
1. In the code.verboo.ai dashboard, click "Create API key" and generate one (prefix vbk_)
2. Export the environment variables (Mac/Linux):
export OPENAI_API_BASE=https://code.verboo.ai/router/v1
export OPENAI_API_KEY=vbk_your_key_here
On Windows:
setx OPENAI_API_BASE https://code.verboo.ai/router/v1
setx OPENAI_API_KEY vbk_your_key_here
3. Check which models your key unlocks:
curl https://code.verboo.ai/router/v1/models \
-H "Authorization: Bearer vbk_your_key_here"
4. Run Aider pointing at the model, with the openai/ prefix:
aider --model openai/glm-5.2
The repo map, auto-commit, and architect mode keep working normally, because they are features of the tool itself, not the provider. Only the model doing the thinking becomes one of the 11 available on Verboo Code.
Frequently asked questions
Can I use Aider with Verboo Code's models?
Yes. Export OPENAI_API_BASE=https://code.verboo.ai/router/v1 and OPENAI_API_KEY with a vbk_ key generated at code.verboo.ai, then run aider --model openai/glm-5.2 (or another model from the list). The repo map, auto-commit, and architect mode keep working normally.
Is Aider really free?
The tool is: open source, Apache 2.0 license, no usage fee. What is not free is the model: you pay directly to the provider you choose, OpenAI, Anthropic, Gemini and others, per token, and that cost can quickly outgrow a fixed-price subscription on long tasks.
Which is cheaper, Aider or Verboo Code?
It depends on volume and model. Aider itself charges nothing, but the token goes to the provider: running Claude Opus 5 through the official API costs US$ 5 and US$ 25 per million input and output tokens. For heavy, recurring agent use that bill grows fast, while Verboo Code charges a fixed price with unlimited tokens starting at R$ 99/month. Compare by the cost of the work actually done at your real volume, not the sticker price alone.
Do I have to give up Aider to use Verboo Code?
No. You can keep Aider as the interface and just swap the engine behind it, pointing OPENAI_API_BASE at Verboo Code's router. Or, if you prefer, Verboo Code itself runs as an independent terminal CLI, in the same command-line format Aider already uses.
If the number that decides is not which model thinks better, but whether the fifth attempt costs the same as the first, Verboo Code solves that with unlimited tokens at a fixed price, whether through its own terminal or by pointing Aider at its engine.
