Novita AI Blog

Unlock the power of AI with the NVIDIA H200. Discover what is H200 and how it enhances generative AI and scientific computing.

Access Kimi‑K2‑Instruct on Novita AI — a trillion-parameter sparse MoE model with 32B activated parameters, 128K context, agentic tool use, and elite performance compared to top mo

Explore the top 5 vision language models for advanced multimodal tasks. Discover their strengths and applications.

Discover GLM-4.1V-9B-Thinking—efficient, powerful, and transparent multimodal reasoning. Explore its groundbreaking performance today!

Learn how to build an MCP server using Novita’s APIs. This guide walks you through the entire process, from creating tools to working with the low level MCP server.

Discover whether MiniMax M1 is free, how to access it, and explore pricing options. Complete guide covering open-source benefits and API costs for developers.

We've upgraded our DeepSeek models to support 160K context length! This major enhancement allows you to handle longer documents, maintain extended conversations, and tackle complex

Explore the differences between L40S vs H100 GPUs and find out which one suits your data center workload requirements.

Discover the hardware requirements for running DeepSeek R1 0528, including GPUs, memory, and storage. Learn how to manage these challenges.

Explore the cost advantages of Llama 3.1 8B and its performance across various tasks with modest hardware requirements.

Learn which models can be run on L40S to maximize your deployment capabilities and manage resources effectively.

Deploy custom Hugging Face models on Novita AI’s LLM Dedicated Endpoint. Enjoy flexible LoRA support, 99.5% SLA, and scalable GPU resources.

Explore which Qwen 3 model suits your needs with its unique blend of accuracy, cost, and memory considerations.