At Brainpool, our mission has always been to provide businesses with the most robust, high-performing, and ethically sound AI solutions available. To achieve this, we built an agnostic AI infrastructure - a framework that allows us to pivot between models based on performance, security, and most importantly, corporate values.

Today, we are announcing a significant shift in our strategy: Brainpool will no longer utilize OpenAI models for our projects.

This decision was not made lightly, but it is one that aligns with our core commitment to excellence and ethical integrity. Here is why we are making the move.

Integrity and the Evolving AI Landscape

One of Brainpool’s foundational values is providing our clients with a choice of ethically driven AI providers. Recent developments at OpenAI have reinforced the shift away from the "safety-first" non-profit ethos that originally defined the organization. More specifically, the recent events that caused our concern are:

The "Department of War" Agreement: OpenAI recently finalized a deal to deploy its models within the classified network of the U.S. Department of War. While OpenAI claims to maintain "human responsibility for the use of force," this partnership marks a stark departure from earlier prohibitions on military applications.
Monetization Over Experience: In a move that prioritizes revenue over user integrity, OpenAI has begun introducing advertisements directly into the chat interface. We believe enterprise AI should be a tool for productivity, not a billboard for sponsored content.
The Safety Exodus: The departure of nearly all co-founders and the mass resignation of the Superalignment (safety) team - including figures like Ilya Sutskever and Jan Leike - has raised serious concerns about the internal priority of AI safety versus "shiny products."

Performance Excellence: The New Leaders

While OpenAI was once the undisputed leader, the landscape has changed. Our internal testing and global benchmarks show that other providers are now delivering superior results for enterprise-grade tasks.

Gemini’s Multimodal Edge: Google’s Gemini 3 series have consistently outperformed GPT models in multimodal reasoning and large-scale data extraction. Its ability to process massive context windows makes it the superior choice for deep-research projects.
Anthropic’s Precision: Anthropic’s Claude Sonnet and Haiku models have set new standards for coding accuracy and reduced hallucination rates. For our clients who require the highest levels of factual reliability and nuanced reasoning, Anthropic is the clear winner.
The Rise of High-Performance Challengers: The gap between traditional giants and emerging leaders is quickly vanishing. Model providers like Mistral and open source options like GLM-5 now rival or exceed GPT models on many benchmarks, giving us a much broader set of options to achieve quali