Live
Google TPUv7 performs excellently in inference benchmarks
Trending storyPast story
Google TPUv7 performs excellently in inference benchmarks
1 reports1 outlets13 days agoUpdate
The story so far
AI overview
Google released performance data for TPUv7 in Qwen 3.5 397B FP8 single inference and the Multi-Model AgentX benchmark, stating that its single inference performance is 50% better than B200 and B300 per dollar and supports ecosystems such as vLLM and SGLang. NVIDIA has reacted strongly and is accelerating its R&D.
AI-written from the reports · Updated 13 days ago
Latest6 Nov, 03:03
Google TPU Stakes Its Claim on Inference with TPUv7 IronwoodCoverage timeline
Follow the coverage to see every side of the story.
6 Nov
- SemiAnalysisGoogle TPU Stakes Its Claim on Inference with TPUv7 Ironwood
Google TPUv7 Ironwood outperforms B200 and B300 by 50% per dollar in single inference performance on Qwen 3.5 397B FP8, and shows competitive or best results in the Multi-Model AgentX benchmark, supporting ecosystems like vLLM and SGLang, but NVIDIA reacts fiercely and accelerates R&D.