Complete rankings of the best AIs for each profession

More than 250 thousand Brazilians already use artificial intelligence at work. But... are you sure you’re using the most effective tool for your field?

The popularization of generative AI has brought models like ChatGPT, Gemini, Claude, LLaMA, and Grok to the forefront of professional productivity. However, many of these systems are evaluated in English, using international data that ignores the context and complexity of Brazilian professions.

And this is where the AI Scoreboard comes in.

Created by the team at Kloel AI, the 2025 AI Scoreboard is the first independent AI benchmark aimed at the Brazilian market, and it answers an essential question:

Which artificial intelligence model is most effective for your profession in Brazil?

Inspired by the real-world challenges faced by Brazilian professionals, the AI Scoreboard evaluates the performance of leading AI tools—such as GPT-4o, Claude, Gemini, LLaMA, and Grok—on practical tasks related to fields such as law, medicine, accounting, Brazilian literature, and general knowledge.

According to Pedro Salles Leite, CEO of Kloel AI:

“We’re living in a time when everyone talks about AI, and many people already use ChatGPT, but few know how to really choose and use the most suitable model. Today, we import this technology from abroad without clearly understanding how it can address the challenges of the Brazilian context.”

In this article, you’ll learn how the AI Scoreboard works, the methodology used, the surprising results, and how it can transform the way you choose and use AI tools in your daily life.

How does the AI Scoreboard work?

Assessment tailored to Brazil

Unlike global benchmarks, the AI Scoreboard is 100% tailored to the Brazilian market. The models are evaluated in Portuguese, using technical and practical content from the country’s main professional fields.

Reliable, transparent, and national methodology

The assessment is based on a rigorous selection of more than 1,000 private, multiple-choice questions, developed from sources such as:

Public service exams (OAB, TCU, etc.)
ENADE and university entrance exams
National laws, technical standards, and guidelines (SUS, Receita Federal)

All models were tested under the same conditions: the same questions, the same language, and no special adjustments or prompt tuning. This ensures a fair and impartial comparison.

The methodology is based on four fundamental pillars:

Usefulness: tests based on real-world use cases in Brazil
Inclusivity: multiple fields and difficulty levels
Reliability: private datasets, audited by experts
Independence: no commercial influence from providers

Models tested

The models evaluated include:

ChatGPT-4o and 4.1

o3 and o4 Mini

Gemini 1.5 Pro and Gemini 3 Pro

Claude 3.7 Sonnet and Claude 3.7 Sonnet Thinking

Grok 3 and Grok 3 Mini

LLaMA 4 Maverick

DeepSeek R1

Qwen QwQ

Mistral Medium

Sabia 3.1

Key results from the 2025 AI Scoreboard

Medicine:

1st place: Gemini 3 Pro (score 8.49)
Claude Thinking performed best on SUS guidelines

Brazilian Literature:

1st place: Claude Thinking (score 8.98)
o3 stood out in questions about Romanticism and Modernism

Accounting:

1st place: o3 (score 9.43)
Gemini performed better on income statements and balance sheets

Legal:

1st place: Gemini 3 Pro (score 9.62)
o3 stood out in Criminal and Constitutional Law

How can you access the full ranking?

The AI Scoreboard is available for free at:

🔗 www.placardaia.com.br

You can filter results by field, compare models, and follow updates as new versions are released.

🔍 Tip: Use the ranking as a strategic tool to choose AIs that truly excel in your field.

Conclusion

Artificial intelligence is already part of the daily routine of Brazilian professionals. But now it's time to take things to the next level: choosing the right AI makes all the difference.

The 2025 AI Scoreboard shows, with transparent data and methodology, which models truly understand Brazil. Use this tool to improve your work and make better-informed decisions.