Qwen 3.8 Flash Next 176B-A6B · Chip comparison
Qwen 3.8 Flash Next 176B-A6B — H200 vs TPU7x
Head-to-head AI inference benchmark comparison of H200 (NVIDIA Hopper) and TPU7x (Google Ironwood) on Qwen 3.8 Flash Next 176B-A6B. 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 →
At 109 tok/s/user interactivity on Qwen 3.8 Flash Next 176B-A6B, H200 delivers 23,466 tok/s/chip at $0.014 per million tokens; TPU7x delivers 30,491 tok/s/chip at $0.009. TPU7x is 35% cheaper per token; TPU7x delivers 30% more tok/s/chip at this point.
H200 posts 12,229 tok/s/chip for $0.028 per million tokens at 148 tok/s/user on Qwen 3.8 Flash Next 176B-A6B; TPU7x posts 23,592 tok/s/chip for $0.012. TPU7x is 56% cheaper per token; TPU7x delivers 93% more tok/s/chip.
Throughput at 187 tok/s/user on Qwen 3.8 Flash Next 176B-A6B: H200 hits 6,893 tok/s/chip, TPU7x hits 18,318. Costs per million tokens land at $0.049 and $0.016 respectively. TPU7x is 68% cheaper per token; TPU7x delivers 166% more tok/s/chip. (Numbers reflect the default agentic-traces · fp8 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | H20023,466TPU7x30,491 | H20012,229TPU7x23,592 | H2006,893TPU7x18,318 |
| Cost ($/M tok) | H200$0.014TPU7x$0.009 | H200$0.028TPU7x$0.012 | H200$0.049TPU7x$0.016 |
| tok/s/MW | H20017,128,281TPU7x25,262,101 | H2008,926,258TPU7x19,546,353 | H2005,031,335TPU7x15,176,401 |
| Concurrency | H200~9TPU7x~22 | H200~8TPU7x~15 | H200~1TPU7x~12 |