Open SourceModelFree

NVIDIA Nemotron 3 Ultra

NVIDIA's open 550B/55B hybrid Mamba-Transformer MoE for long-running, high-throughput agents, shipped with weights, data, and training recipes. vLLM day-0; NVFP4 runs on Hopper and Blackwell.

Homepage ↗

Alternatives

DeepSeek V4MiniMax M3GLM-5.1

1 workflow use NVIDIA Nemotron 3 Ultra

Local InferenceFreeMachine-verified

Serve NVIDIA Nemotron 3 Ultra yourself for high-throughput agents (vLLM)

Stand up the NVFP4 Nemotron 3 Ultra checkpoint as an OpenAI-compatible endpoint for fast, long-running agent loops, validated serve flags + endpoint.

vLLMNVIDIA Nemotron 3 Ultra· advanced