How they use Nebius

Zencoder uses Nebius AI Studio for open-model inference behind its coding assistant. Developer advocate Leon Malisov describes hosted models for code completion and help selecting, testing and integrating them. Zencoder also uses other model providers. Nebius’s August 2025 customer case describes the arrangement.

The case study also describes multi-node training on Nebius AI infrastructure to optimize and fine-tune models for software engineering tasks. AI Studio supplies inference, while Nebius provides model-selection and optimization guidance, early model access and architectural consulting. A September 2025 Nebius post reports 30M+ accepted code-completion characters and 200 ms p50 latency for the AI Studio inference response. The same post separately describes consistent sub-100 ms inference latency and rapid, flexible scaling without infrastructure overhead. Nebius also reports improved product reliability, customer adoption and reduced operational complexity for Zencoder.

The case study also describes continued use of AI Studio’s expanding model portfolio to develop next-generation autonomous coding agents. Planned tasks include technical-debt management, proactive bug detection and end-to-end testing automation.