Intelligence that speaks
your language

Read the documentation Go to API platform

Connect your system to our specialized AI models

Maritaca's API offers a simple, secure and scalable way to integrate our language models, including the entire Sabiá family, directly into your applications. Build smarter, more efficient solutions connected to Brazilian and global context.

Explore our platform, access full technical documentation, and start expanding your applications with the most advanced national AI today.

Price and performance

Frontier-level quality at about a third of the cost — compared on the same benchmarks.

Cost to run the suite Quality
Sabiá 4 Thinking
R$ 206
90.8%
Gemini 3.1 Pro
R$ 281
92.4%
GPT-5.4
R$ 449
92.8%
Opus 4.8
R$ 590
92.5%

Total cost to run the full benchmark suite · quality = overall average accuracy · GPT, Opus and Gemini converted at R$ 5.14/US$.

Meet our models

Sabiá 4 Thinking

Price per 1M tokens:

Input: R$ 5.00 / Output: R$ 40.00

  • • 128,000 tokens of context
  • • Our reasoning model
  • • Optional inference 100% in Brazil
Get started

Sabiá - 4

Price per 1M tokens:

Input: R$ 5.00 / Output: R$ 20.00

  • • 128,000 tokens of context
  • • Our most advanced model
  • • Optional inference 100% in Brazil
Get started

Sabiazinho - 4

Price per 1M tokens:

Input: R$ 1.00 / Output: R$ 4.00

  • • 128,000 tokens of context
  • • Our fastest model
  • • Optional inference 100% in Brazil
Get started

For inference 100% on national territory, call the same model with the -br-sp suffix (for example, sabia-4-thinking-br-sp). Same capabilities, 30% above the standard price. Learn more.

Plans for companies

Enterprise Plan

For companies needing a contract, dedicated SLA, custom billing or priority support.