以下部分引用 AI 总结,现阶段 AI 仍然有幻觉。请以内容的原文为准。
数据新鲜度群聊 07-14 ✓卖方 07-14 ✓Wrap 断档15天 ✕News 07-15 ✓X 断档12天 ✕页面生成 07-15
NVIDIA GB300 NVL72
6 atoms · 跨 3 天 · 首见 2026-06-16 · 最近 2026-06-30
三色: 🟦 fact 5 · 🟥 take 1 · stance ▲5/▼0/◆0
来源: X 3 · 卖 3
时态: fresh:6
标签: 好数字:5 · 好观点:3 · 好信源:1
叙事 (narrative) (1)
- 2026-06-29 · inference performance claim
For enterprises building the next generation of agentic AI, this means more powerful autonomous and domain-specific agents, backed by the inference performance and lowest cost per token.
tags: 好观点 · → daily
事实 (fact) (5)
- 2026-06-29 ⭐ · deployment with Microsoft
📣 @AnthropicAI Claude models are now generally available in @Microsoft Foundry, running on the NVIDIA GB300 NVL72 platform on Azure with NVIDIA Quantum-X800 InfiniBand networking.
tags: 好数字·好信源·好观点 · → daily
- 2026-06-16 · Agentic Coding Performance vs H200
NVIDIA GB300 NVL72 delivers 20x better agentic coding performance compared to H200.
tags: 好数字·好观点 · → daily
value: qty=20x · date=2026-06-16
- 2026-06-30 · 平均功耗
Blackwell GB300 NVL72 reaches 134–140 kW
tags: 好数字 · → daily
value: qty=134-140 · unit=kW
- 2026-06-16 · Concurrent Agents per MW at 20 Tokens/sec SLO
NVIDIA GB300 NVL72 supports over 60,000 concurrent agents per MW at a Service Level Objective of 20 tokens per second.
tags: 好数字 · → daily
value: qty=>60,000 · date=2026-06-16
- 2026-06-16 · Concurrent Agents per MW at 60 Tokens/sec SLO
NVIDIA GB300 NVL72 supports over 20,000 concurrent agents per MW at a Service Level Objective of 60 tokens per second.
tags: 好数字 · → daily
value: qty=>20,000 · date=2026-06-16
时间轴 (近 20)
- 2026-06-30 ·
fact · 平均功耗 · → daily
- 2026-06-29 ·
fact · deployment with Microsoft · → daily
- 2026-06-29 ·
narrative · inference performance claim · → daily
- 2026-06-16 ·
fact · Agentic Coding Performance vs H200 · → daily
- 2026-06-16 ·
fact · Concurrent Agents per MW at 20 Tokens/sec SLO · → daily
- 2026-06-16 ·
fact · Concurrent Agents per MW at 60 Tokens/sec SLO · → daily
← 实体目录 · 系统日志