Share your ideas about our platform
Qwen3-Coder-480B-A35B-Instruct
Are there any plans to add the Qwen3-Coder-480B-A35B-Instruct? https://huggingface.co/Qwen/Qwen3-Coder-480B-A35B-Instruct
Nebius AI Studio App (for Android and iOS)
I would like to suggest developing dedicated Android and iOS applications for Nebius AI Studio. Currently, the platform’s usability on mobile devices is very limited, the non-responsive design makes it difficult to interact with input fields and use key features efficiently. Enhancing mobile device support, either through a more responsive web design or native mobile apps, would significantly improve the user experience. It would also encourage more frequent use of Nebius AI Studio, especially for testing different language models and prompts.
Feature Request: Add Cross-Encoder Reranking Support via /rerank or /score Endpoint
It would be great to have native support for reranking in the Nebius AI API, especially for use in retrieval-augmented generation (RAG) or semantic search workflows. See https://docs.vllm.ai/en/latest/serving/openai_compatible_server.html for the /rerank and /score API. Add a dedicated endpoint for reranking such as the /score or the /rerank endpoind. Add cross-encoder models to the available model library
Gestionale palestra
Creare app gestione da Palestra Ispirarsi a Xeniasoft.it La mia palestra è www.centrocami.it Professionale Ho 18 collaboratori e 300 clienti da gestire. Sono una A.S.D.
Feature Request: Support for QLoRA Fine-Tuning
Currently, Nebius AI Studio supports both LoRA fine-tuning and full fine-tuning approaches. Could you please add support for QLoRA (Quantized LoRA) fine-tuning, which combines quantization with LoRA for more memory-efficient training?
Qwen3 for single model use (Disable Thinking)
The Qwen3 series are capable of “Uniquely support of seamless switching between thinking mode (for complex logical reasoning, math, and coding) and non-thinking mode (for efficient, general-purpose dialogue) within single model, ensuring optimal performance across various scenarios.” Please make a new models for the FAST and use of within single model, ensuring optimal performance across various scenarios.
Sciance and Technology
Nepal is culturally diverse country it holds key cultures, maithali art culture is one of them this art culture contain various science relate paintings which can spread reserch for education . So I through my freelance company with design and logo help promoting education.
could you run docling?
https://github.com/docling-project/docling is a great document extractor. It is able to extract data from all kinds of documents and you can set it up to also do OCR. Would be great to have this for all kinds of AI solutions.
With millions of stolen credentials, and an estimated of 10 000 billions of damage in 2025, i will not provide my credit card to any organization on internet, but a one time virtual card.
I did not understood - nor test, how i can use a virtual credit card, to add pre paid credit to my account ? Is it possible ? Will it be implemented ? I tried to contact the support but it is a dead end…
Studio and Cloud
Make it easy to move between the Nebius Cloud and Nebius Studio and work from a common pool of money.