OpenAI Opens in Brazil: What Changes for Developers
OpenAI now has commercial operations in São Paulo. Separate the local team and developer programs from adoption claims, product access, and data residency.
OpenAI launched commercial operations in Brazil on August 27, with a local team based in São Paulo. For developers, the immediate change is local commercial and community support around products that were already available—not a new Brazil-only model, API endpoint, price, or data-residency region.
The company announcement combines four different developments: the São Paulo operation, developer events, institutional programs, and OpenAI’s own adoption statistics. Those developments can reinforce one another, but they are not interchangeable evidence. An office can make sales and support closer to customers; a workshop can help builders learn a product; a memorandum can begin a public-sector project. None alone proves production adoption or changes the technical terms of an API deployment.
Reuters independently confirmed the São Paulo office and commercial launch. Its report also preserved the most important qualifier on the scale story: OpenAI did not disclose a Brazilian user count.
What OpenAI actually launched in Brazil
The confirmed change is a locally based commercial operation intended to work with Brazilian companies, developers, researchers, and public institutions. OpenAI also described several related activities, but their delivery status varies.
| Announcement element | What is confirmed | What it does not establish |
|---|---|---|
| São Paulo operation | A local team and commercial presence | A new model, local API endpoint, or Brazil-specific price |
| OpenAI para Devs | A Portuguese-language roadshow reached 285 participants in Recife, Rio de Janeiro, Florianópolis, and São Paulo | Broad adoption across Brazil’s developer population |
| ITA partnership | ChatGPT Edu accounts and credits are planned for students, faculty, and staff | The number of accounts, deployment date, or measured learning outcome |
| Research support | IMPA and HCFMUSP projects are named under OpenAI’s grants program | Completed research results or validated public-service systems |
| São Paulo and Prodam | A memorandum of understanding was signed | A production deployment, procurement award, or approval for a specific use case |
The distinction is especially important for the public-sector language. A memorandum of understanding records an intended collaboration. The announcement says OpenAI and São Paulo’s municipal technology company, Prodam, want to make AI available to the city and other public administrations; it does not identify a production system, contract value, data flow, launch date, or acceptance test.
The developer program is more concrete but narrower. OpenAI says its first OpenAI para Devs roadshow ran entirely in Portuguese and included technical presentations, Codex demonstrations, workshops, and local partners. It ended with a São Paulo hackathon whose winning prototype used public data to help small businesses find procurement opportunities. That demonstrates that an event and prototype occurred. It is not evidence that the prototype entered service or that similar projects will clear procurement, security, accessibility, and operational review.
API access already existed; local residency did not appear with the office
Brazil appears on OpenAI’s current list of countries supported for API access, and the launch announcement does not describe API access as newly enabled. The commercial operation therefore should not be read as a new technical availability date.
The same caution applies to infrastructure. OpenAI’s current API data-control documentation lists the regions available for configured storage and processing. Brazil is not among the listed residency regions as of August 28. Opening an office in São Paulo does not, by itself, say where customer content is stored or processed.
That leaves a practical separation for a Brazilian team evaluating the announcement:
- commercial locality: sales, account relationships, partnerships, and support may now involve a team in Brazil;
- service availability: the API was already supported in Brazil, while product and feature access still depends on the relevant plan and documentation;
- technical locality: a local office is not a regional API endpoint or a promise of in-country storage and inference; and
- deployment compliance: each organization still has to map its data, vendors, contracts, and use case to its own legal and governance requirements.
This is not a finding that a particular deployment violates Brazilian law. It is a reminder that corporate presence and technical data location answer different questions.
The adoption numbers are useful signals with missing denominators
OpenAI says Brazil is one of ChatGPT’s three largest markets by weekly active users, that its Brazilian user base nearly doubled over the past year, and that people in Brazil send about 215 million ChatGPT messages a day. Reuters reported those figures as company statements and noted that OpenAI withheld the underlying user count.
The announcement adds more developer-specific claims: Brazil ranks second by the number of developers using the OpenAI API, is Codex’s largest Latin American market, and has seen weekly Codex users grow more than elevenfold since the start of 2026. It also says daily Codex interactions rose nearly thirtyfold.
Those numbers show the shape of OpenAI’s case for investing locally. They do not reveal the starting counts, the definition of a developer, retention, paid conversion, task success, or whether growth came from sustained production use rather than a newer product’s launch curve. Message volume is activity, not a user count, and neither measure demonstrates economic impact.
The company also reports that 35% of classified messages from individual Brazilian ChatGPT accounts were work-related in June, compared with 30% globally. Of that work-related subset, 53% asked ChatGPT to complete a task or produce an output. The announcement does not publish the classification method, sample size, uncertainty, or a country-level comparison over time. The figures are informative vendor telemetry, not an independent survey of Brazilian workers or developers.
What would show that the developer push changed practice
The strongest near-term evidence will come from delivery details rather than another usage multiple. Developers and technical leaders can watch for:
- published dates, eligibility, account counts, and evaluation plans for the ITA and research programs;
- durable Portuguese-language documentation, support paths, events, and community programs beyond the launch week;
- production follow-through from the hackathon or other locally designed applications;
- specific scope, procurement, governance, and outcome reporting for public-sector work;
- independently described case studies that include costs, failure modes, human review, and measurable results; and
- any documented change to regional product availability, billing, contracting, support, or data controls.
The local operation can matter without changing the API itself. A team that understands Brazilian organizations, works in Portuguese, and can maintain relationships after a launch event may reduce the distance between a global product and local deployment. But that value has to appear in continued support, shipped programs, and observable outcomes.
For developers, the accurate reading is therefore modest and concrete: OpenAI has put commercial staff and community activity closer to a large existing market. The office makes more local work possible. The programs and vendor metrics do not yet prove how much of that possibility will become reliable production use.