Brazilian AI
Rogue, Adapta or Maritaca: which one?
Updated · Rogue
Short answer
They are three different offers. Maritaca AI trains language models for Portuguese (the Sabiá family). Adapta is a hub: a subscription that brings together other companies' AI models, with courses. Rogue researches agents and delivers projects: websites, web apps, slides and games, with results published on global leaderboards.
| Criterion | Rogue | Adapta | Maritaca AI |
|---|---|---|---|
| What it is | Agent research lab and product | AI hub: one subscription to other companies' models, with courses | Language models for Portuguese |
| What you get | Published websites, web apps, slides and games; Rogue Code for Mac | Multi-model chat (ONE), no-code builder (Skip), immersions | Sabiá 4, Sabiazinho 4, Sabiá 4 Thinking and S5, through an API |
| Best for | Building and publishing finished work | Chatting with several models and learning | Portuguese and legal text through an API |
Sources checked on November 6, 2026.
When to choose each one
Choose Maritaca AI if you need a model through an API for Portuguese text, especially legal text. Choose Adapta if you want a hub, one subscription to chat with other companies' models and learn from courses. Choose Rogue if you want the AI to build and publish the work, with the research behind it measured in public.
Published research
Three results from 2026, each with its own article, data and evaluation notes.
Terminal-Bench 4.0October 2026
71.21% Rogue Constellation
Our flagship coding agent. Ahead of Claude Opus 5.5 (66.4%) and GPT-6 Astra (57.9%); 71.43% on Terminal-Bench-Science 0.1.
Read the research
DeepSWE v1.1September 2026
75.22% Rogue DeepThink Preview
Our strongest code writer yet, built on GLM 5.3 and only on Rogue: 6.19 points above its own base on DeepSWE v1.1, ahead of GPT-6 Astra (74.12%).
Read the researchTerminal-Bench 4.0November 2026
−15% cost Effort router
A model trained by Rogue reads the task before the agent starts and sets its reasoning effort. Same pass rate as high effort at 15% lower cost; 771,072 parameters trained in under an hour.
Read the researchQuestions
Which one publishes results on international agent benchmarks?
Rogue publishes results on Terminal-Bench 4.0, Terminal-Bench-Science 0.1 and DeepSWE v1.1. Maritaca AI publishes technical reports on its Sabiá models, with evaluations in Portuguese.
Which one builds websites and apps?
Rogue builds and publishes websites, web apps, slides and games. Adapta offers Skip, a no-code software builder. Maritaca AI offers language models through an API.
Can I start for free?
On Rogue, yes: it is free during the beta, with usage limits. For the others, check each company's pricing page.