DownloadNVIDIA Nemotron 3.5 Bolt 74B-A9B 24GB is an open-weight multimodal AI model optimized for running autonomous agents on systems with 24GB of GPU memory. Derived from Nemotron 3.5 Super VL through NVIDIA's Star Elastic compression method, it features 74B total parameters with approximately 9B activated per token. Its hybrid Mamba-2–Transformer LatentMoE architecture incorporates a C-RADIO v4-H vision encoder for processing text, images, and video. The model supports a 256K-token context window and uses IQ2 quantization with quantization-aware distillation to reduce memory requirements while preserving reasoning capabilities. Designed for local coding assistants, browser automation, terminal operations, and tool calling, it supports configurable reasoning modes for balancing speed and accuracy. LiLiCorr speculative decoding accelerates text generation, while vLLM integration enables deployment on compatible NVIDIA GPU systems.
Features
- 74B total parameters with approximately 9B active per token
- Hybrid Mamba-2–Transformer LatentMoE architecture
- C-RADIO v4-H vision encoder for image and video understanding
- 256K-token context window for extended agent workflows
- IQ2 quantization optimized for 24GB GPU memory
- Star Elastic compression with quantization-aware distillation
- Configurable reasoning modes and LiLiCorr speculative decoding
- Local deployment through vLLM for coding, browsing, and tool use
Project Samples
