Head-to-head AI inference benchmark comparison of GB300 NVL72 (NVIDIA Blackwell) and TPU7x (Google Ironwood) on Kimi K3 2.8T. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics β same interactions as the main inference chart.
AgentX replays real coding-agent sessions rather than fixed-length prompts, so context grows turn over turn and most of each request is served from cache instead of being recomputed. That turns the comparison into a systems question: KV transfer between nodes, prefix-aware routing, and cache capacity all move the curve alongside raw chip throughput. Fixed-sequence workloads stay the clean baseline for kernel and silicon performance, so the two scenarios answer different questions about the same hardware. Learn more about AgentX β
Near the low end of the 26β390 tok/s/user interactivity band, at 116 tok/s/user on Kimi K3 2.8T: GB300 NVL72 runs 7955 tok/s/chip at $0.08/M tokens, TPU7x runs 4014 at $0.07/M. TPU7x is 13% cheaper per token; GB300 NVL72 delivers 98% more tok/s/chip.
Setting 208 tok/s/user as the target on Kimi K3 2.8T, GB300 NVL72 produces 4054 tok/s/chip ($0.16 per million tokens) and TPU7x produces 1480 ($0.19). GB300 NVL72 is 22% cheaper per token; GB300 NVL72 delivers 174% more tok/s/chip.
At 299 tok/s/user interactivity on Kimi K3 2.8T, GB300 NVL72 delivers 2124 tok/s/chip at $0.30 per million tokens; TPU7x delivers 941 tok/s/chip at $0.30. Cost per token is essentially tied; GB300 NVL72 delivers 126% more tok/s/chip at this point. (Numbers reflect the default agentic-traces Β· fp4 selection for this URL β table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB300 NVL72:7954.6TPU7x:4014.3 | GB300 NVL72:4054.2TPU7x:1479.7 | GB300 NVL72:2124.3TPU7x:941.1 |
| Cost ($/M tok) | GB300 NVL72:$0.081TPU7x:$0.071 | GB300 NVL72:$0.158TPU7x:$0.193 | GB300 NVL72:$0.302TPU7x:$0.304 |
| tok/s/MW | GB300 NVL72:3752189TPU7x:3325856 | GB300 NVL72:1912382TPU7x:1225899 | GB300 NVL72:1002045TPU7x:779661 |
| Concurrency | GB300 NVL72:~47TPU7x:~21 | GB300 NVL72:~4TPU7x:~6 | GB300 NVL72:~1TPU7x:~3 |
Agentic inference metrics from the AgentX scenario and fixed-sequence inference metrics across models, hardware configurations, and serving parameters.
No measurements to plot for this selection. Review the benchmark controls above or adjust quick filters.
