Novita AI is a serverless GPU cloud platform providing fast, cost-efficient API access to a wide catalog of open-source AI models. It offers OpenAI-compatible endpoints, flexible auto-scaling infrastructure, and pay-as-you-go pricing optimized for producti
Novita AI serverless inference powered by Kimi K3 — Moonshot AI's most capable model with 2.8 trillion parameters, built on Kimi Delta Attention with a 1M-token context window. Supports native visual understanding, function calling, and structured output. ...
Ling 3.0 Flash Sante is a health and medicine-enhanced MoE model from InclusionAI with 124B total parameters and 5.1B activated per token. Optimized for medical knowledge reasoning, clinical safety, and evidence-based retrieval. Supports function calling w...
Ling 3.0 Flash Fin is a finance-enhanced MoE model from InclusionAI with 124B total parameters and 5.1B activated per token. Optimized for investment workflows, multi-step reasoning, and long-horizon planning. Supports function calling with a 256K token co...
DeepSeek V4.1 Flash is a reasoning model from DeepSeek optimized for fast inference. It supports text and image input with function calling and structured output. The model handles a 1M token context window with up to 384K output tokens.
DeepSeek V4 Pro 0813 is a reasoning model from DeepSeek with exceptional reasoning performance. Text-only input with 1M context window and 384K max output. Supports function calling and structured output.
MiniMax M3 is a multimodal reasoning model from MiniMax supporting text, image, and video input with function calling and structured output. It handles a 977K token context window with tiered pricing based on input length.
GLM-5.3 is a reasoning model from Z.ai that thinks through problems step by step before answering. It supports function calling and structured output, and handles long documents with its 1M token context window and 128K token output capacity.
GLM-5.3 Flash is a native multimodal model from Z.ai supporting text, image, and video input with a 1M context window. It uses a hybrid sparse and linear attention architecture for efficient long-context processing at a fraction of GLM-5.3's cost.