Unverified50% confidenceEventExact time
NVIDIA的Nemotron 3.5 Lightning模型经过2-bit量化后仍能持续运行工具调用长达10分钟以上
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/24/2026
First Seen
Valid until: 11/22/2026
Sources
Related Entities
Related Claims
UnverifiedNemotron 3.5 Lightning is equipped with D-Flash speculative decoding technology to accelerate generation, and its weights, data, and training recipe are all open.73% similarUnverifiedNemotron 3.5 Lightning is a 30B-scale Mixture of Experts (MoE) model that activates only approximately 3B parameters per inference.67% similarUnverifiedNemotron 3.5 Lightning is reportedly distilled from the larger Nemotron 3 Ultra.66% similarUnverifiedNVIDIA has developed a model called Nemotron Ultra with 550 billion parameters65% similarUnverifiedNVIDIA NIM内置了TensorRT-LLM等推理加速引擎65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/795643API
curl https://kongchang.com/api/v1/knowledge/claims/795643MCP
get_claim(id=795643)