An independent developer has successfully modified Google’s open-source Gemma 4 AI model to replicate the capabilities of Anthropic’s advanced Claude Opus system, creating a new family of locally deployable AI tools called Gemopus. The project, led by developer Jackrong of Qwopus, demonstrates how open-source AI models can be adapted to match proprietary systems’ performance.
Google released Gemma in February 2024 as its first major open-source AI model family, positioning it as a more accessible alternative to closed systems like Gemini. Anthropic’s Claude Opus, released in March 2024, represents the company’s most powerful commercial AI offering to date. The Gemopus project bridges these two approaches by applying fine-tuning techniques to make Gemma behave like Claude Opus.
‘This shows the flexibility of open-weight models,’ said an AI researcher familiar with the project who requested anonymity. ‘With the right adjustments, smaller models can approximate much larger systems’ behavior.’ The modification could make high-quality AI more accessible for users with limited hardware, as Gemma’s smaller size allows it to run on less powerful devices.
Industry analysts note the development highlights growing competition in the AI space, where open-source alternatives increasingly challenge proprietary models. ‘We’re seeing a Cambrian explosion of model variants,’ said a tech analyst at a major research firm. ‘The barrier to creating competitive AI systems is lowering.’
The project raises questions about intellectual property boundaries in AI development and whether fine-tuned models could face legal challenges. Some experts suggest such adaptations might accelerate innovation, while others warn they could discourage commercial investment in cutting-edge AI research.