DeepSeek's low-cost open AI models are challenging the dominance of expensive proprietary systems, forcing enterprises to reconsider whether they actually need premium AI services. In late 2024, the Chinese firm DeepSeek released V3, an open model reportedly trained at a fraction of the cost that US frontier labs spend on their systems. A reasoning model called R1 followed in January 2025 and was seen as a legitimate threat to frontier models from companies like OpenAI, Anthropic, and Google.

The shift reflects a broader industry trend. According to research firm SemiAnalysis, "with each generation, open-source models take half as long to catch up to the first closed-source model of the era." This acceleration means that the gap between cutting-edge proprietary AI and freely available alternatives is shrinking faster than many expected.

What's the Difference Between Open-Weight and Open-Source Models?

Understanding the landscape of "open" AI models is crucial for enterprises deciding which approach fits their needs. The term "open" can mean different things, and those differences matter significantly for how companies can use and customize these systems.

The most common open models used in enterprises are open-weight models. These allow companies to customize AI services and tools to their internal requirements. Corporate leaders and IT decision-makers can see inside the model, audit it, and tune it to their own specific data. Open-weights are the parameters processed by mathematical techniques to produce an output. Enterprises can customize a model by fine-tuning the weights, adding their internal data, and deploying it in-house.

However, open-weight models hide information such as code and training data, so some parts cannot be modified. According to the Open Source Initiative (OSI), a truly open-source model also releases the data it was trained on, along with other information allowing those models to be studied, inspected, used, modified and freely distributed. This distinction matters because it determines how much control an enterprise actually has over the system.

"Open-weight models help provide that assurance by allowing organizations to control their own data, evaluate and adapt models to their own needs, and deploy them wherever their business requirements demand," said Jensen Huang, CEO of Nvidia.

Jensen Huang, CEO at Nvidia

Why Are Enterprises Moving Away From Expensive Proprietary Models?

While ChatGPT, Google's Gemini, and Anthropic's Claude provide well-rounded AI capabilities, they come with significant costs and may be overkill for specific corporate uses. Many enterprises can be better served by a smaller language model (SLM) or large language model (LLM) that is focused on their specific needs. Open models function as blank canvases on which enterprises can paint their workflows.

The practical reality is that most organizations do not need general-purpose AI trained on the entire internet. Deepak Seth, senior director analyst at Gartner, explained this mismatch: "An enterprise's real needs sit in specific workflows with specific data, and a general-purpose closed model trained on the entire internet is overkill for most of them." This insight explains why companies are increasingly exploring alternatives to premium proprietary systems.

Recent deployments demonstrate this shift in action. Companies such as ServiceNow and RWS have deployed dozens of open-source models in addition to proprietary models, each specializing in specific tasks. Max Goss, research director at Gartner, noted that enterprises should not fear adopting a multi-vendor approach: "We shouldn't be afraid to adopt a multi-vendor approach if we think that we can get value from different AI tools rather than risk the lock-in of having a single AI tool".

How to Evaluate Open Models for Your Enterprise

Organizations considering a shift toward open models should evaluate several key factors to ensure they select the right approach for their specific needs and constraints.

• Data Control and Security: Open models can be run internally and cut off from the cloud, providing additional security especially for regulated industries where data control is paramount. This on-premise deployment option is critical for organizations handling sensitive information.
• Operational Efficiency: For physical operations such as in vehicles or on-site, decisions need to happen in milliseconds, sometimes directly on the device. Purpose-built, efficient models that run close to where data is generated are essential for real-time applications.
• Governance and Auditability: Open models allow enterprises to inspect the weights, audit the training data, or air-gap the deployment. As one analyst noted, "You can't govern what you can't see," making transparency a critical advantage.
• Customization for Regional Needs: Open models can help countries and enterprises customize AI to meet indigenous customs, traditions, policies and regional regulatory constraints, supporting what experts call "sovereign AI."

Praveen Murugesan, vice president of engineering at Samsara, explained how open models enable layered deployment: "Open models make that layered deployment possible because the enterprise controls where each model runs and what data it touches".

What Are the Risks of Moving to Open Models?

Despite their advantages, open models introduce new challenges that enterprises must carefully manage. Deployment, maintenance and updates in many cases fall in the hands of enterprises rather than vendors, creating operational burden. Additionally, open models might not always be completely vetted, introducing a level of risk that proprietary information in a model could be leaked beyond the borders of an enterprise.

There is also a tension between openness and security. Craig LeClair, vice president and principal analyst at Forrester Research, observed: "Open source models will be run in controlled on-premise environments, which just makes them less open source pretty quickly." Companies wanting models trained on their proprietary information will need solid control over the tools because of intellectual property leakage worries.

Jinsook Han, founder and partner at Spruce Peak Ventures, emphasized the importance of building governance into open model deployments: "You actually build the boundaries around it. So the responsible AI is built in".

How Is the Global AI Landscape Shifting?

The rise of open models like DeepSeek's offerings reflects a geopolitical shift in AI development. China is a proponent of open source and open weight models as it increasingly becomes an AI rival to the US. Other major economies, including Germany, France and India, are also encouraging the adoption of open models.

Richard Morton, vice president and managing director at the Abu Dhabi-based Institute of Foundation Models at Mohamed bin Zayed University of Artificial Intelligence, stressed the importance of open models for national interests: "Open models are important to sovereign AI so nations can understand, adapt, and control systems powering digital infrastructure".

The innovation advantage also favors open approaches. Kari Briski, Nvidia's vice president of generative AI software, noted that open models do a better job of innovating based on localized knowledge: "Open models and open data are that bootstrap: you don't have to recreate capturing the knowledge of the internet as a pre-training model".

Kari Briski, Nvidia's vice president of generative AI software

As enterprises continue to evaluate their AI strategies, the emergence of cost-effective open models like DeepSeek's V3 and R1 is forcing a fundamental reckoning. The question is no longer whether open models can compete with proprietary systems, but rather which combination of open and proprietary tools best serves an organization's specific needs. Samar Abbas, co-founder and CEO of Temporal, captured the emerging consensus: "We are a big believer in open source and we are super excited to see the ecosystem around these open-source models build and thrive. In the fullness of time, these open-source models will have their own space".