GLM 5.3 744B — GB200 NVL72 vs TPU7x
Head-to-head AI inference benchmark comparison of GB200 NVL72 (NVIDIA Blackwell) and TPU7x (Google Ironwood) on GLM 5.3 744B. Latency, throughput, and cost across LLM workloads. Use the chart controls below to switch sequences, precisions, and metrics — 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 →
GB200 NVL72 hits 16496 tok/s/chip for $0.03 per million tokens at 154 tok/s/user on GLM 5.3 744B. No TPU7x data at this operating point.
GB200 NVL72: 12967 tok/s/chip, $0.04 per million tokens at 212 tok/s/user on GLM 5.3 744B. TPU7x is unmeasured here.
At 271 tok/s/user on GLM 5.3 744B, GB200 NVL72 delivers 6601 tok/s/chip at $0.08 per million tokens; TPU7x hasn't been benchmarked at this target. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | GB200 NVL72:16496.4TPU7x:— | GB200 NVL72:12967.4TPU7x:— | GB200 NVL72:6600.7TPU7x:— |
| Cost ($/M tok) | GB200 NVL72:$0.031TPU7x:— | GB200 NVL72:$0.040TPU7x:— | GB200 NVL72:$0.078TPU7x:— |
| tok/s/MW | GB200 NVL72:8821601TPU7x:— | GB200 NVL72:6934465TPU7x:— | GB200 NVL72:3529763TPU7x:— |
| Concurrency | GB200 NVL72:~58TPU7x:— | GB200 NVL72:~45TPU7x:— | GB200 NVL72:~17TPU7x:— |