End of Free Access to DeepSeek V4 Flash on NVIDIA Build

NVIDIA has discontinued DeepSeek V4 Flash (still in development) from its NVIDIA Build platform servers, effective today. This means those who accessed this model for free will no longer be able to do so. For those unfamiliar with the platform: NVIDIA Build is a service that offers free APIs for AI models, both proprietary (e.g., Nemotron) and third-party (Llama, Qwen, GPT-OSS, among others). It is designed for testing and prototyping, with rate limits, not for production. If a prototype works well, the ideal next step is to migrate to paid options like NVIDIA NIM or a cloud provider. DeepSeek was one of the most popular models on the platform, and its removal particularly affects students and people learning about AI who cannot yet afford paid services. It's worth noting that DeepSeek is one of the most affordable models to access through its own platform or third-party providers. Screenshots: Proof that the model was previously available on NVIDIA Build Message: model discontinued Home screen showing the most popular models. Previously, DeepSeek appeared among the top results. A project failure (personal project) that was using the API for testing, which would later be migrated to DeepSeek's own API