Model

Llama-3.3 Nemotron Super 49B v1

NVIDIA

开源 Llama 3.1 Community License

Llama-3.3-Nemotron-Super-49B-v1 is a large language model (LLM) derived from Meta Llama-3.3-70B-Instruct. It's post-trained for reasoning, chat, RAG, and tool calling, offering a balance between accuracy and efficiency (optimized for single H100). It underwent multi-phase post-training including SFT and RL (RLOO, RPO).

发布日期2025年3月18日

参数规模49.9B

上下文长度—

许可证Llama 3.1 Community License

知识截止2023年12月31日

论文模型权重官方博客在线体验对比此模型

Benchmarks

评测成绩

评测基准	类别	分数	来源
MATH-500	math reasoning	96.6	来源
MT-Bench	communication reasoning general roleplay	91.7	来源
MBPP	reasoning general	91.3	来源
Arena Hard	general reasoning creativity	88.3	来源
BFCL v2	general reasoning	73.7	来源
GPQA	reasoning general	66.7	来源
AIME 2025	math reasoning	58.4	来源

Pricing

API 价格对比

暂无 API 价格。