Event ·Multi-source

NVIDIA Nemotron 3.5 Transcribe 语音识别模型开源，vLLM 提供 day-0 支持

NVIDIA 发布开源语音识别模型 Nemotron 3.5 Transcribe，支持 40 种语言，可处理最长一小时的录音。模型支持通过提示词定制转录内容，并输出说话人标签和音频问答功能。开源权重提供 BF16、FP8 和 NVFP4 三种精度格式，vLLM 与 NeMo Speech 在发布当天即可部署服务。

Current read

NVIDIA 开源了新的语音识别模型，支持 40 种语言和一小时长音频，还能标说话人、做音频问答，用 vLLM 当天就能跑起来。

11 sources62° AI heatLast updated 2026/10/28 15:26:59

Evidence

11 sources

Primary sourcevLLM

2026/10/28 15:26:59View source

Related source 1Artificial Analysis

2026/10/28 15:05:14View source

Related source 2Hugging Face: Blog

2026/10/28 14:58:17View source

Related source 3Decoder

2026/10/28 15:49:39View source

Related source 4techcrunch

2026/10/28 15:00:00View source

Related source 5rohanpaul_ai

2026/10/27 16:41:33View source

Related source 6CoreWeave

2026/10/26 17:38:48View source

Related source 7TestingCatalog

2026/10/28 09:58:40View source

Related source 8Gorden Sun

2026/10/28 16:01:32View source

Related source 9arXiv cs.AI

2026/10/27 08:46:21View source

Related source 10面壁智能 ModelBest

2026/10/27 11:20:48View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story