Qwen 3.8 Flash Next 176B-A6B · Chip comparison
Qwen 3.8 Flash Next 176B-A6B — MI355X vs Vera Rubin NVL72
Head-to-head AI inference benchmark comparison of MI355X (AMD CDNA 4) and Vera Rubin NVL72 (NVIDIA Vera Rubin) 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 →
Throughput at 166 tok/s/user on Qwen 3.8 Flash Next 176B-A6B: MI355X hits 37,441 tok/s/chip, Vera Rubin NVL72 hits 491,079. Costs per million tokens land at $0.011 and $0.002 respectively. Vera Rubin NVL72 is 82% cheaper per token; Vera Rubin NVL72 delivers 13× the tok/s/chip.
MI355X / Vera Rubin NVL72 on Qwen 3.8 Flash Next 176B-A6B at 241 tok/s/user: 24,619 / 212,232 tok/s/chip, $0.017 / $0.005 per million tokens. Vera Rubin NVL72 is 72% cheaper per token; Vera Rubin NVL72 delivers 8.6× the tok/s/chip.
Toward the upper edge of the 92–389 tok/s/user interactivity band, at 315 tok/s/user on Qwen 3.8 Flash Next 176B-A6B: MI355X runs 20,726 tok/s/chip at $0.020 per million tokens, Vera Rubin NVL72 runs 195,094 at $0.005. Vera Rubin NVL72 is 74% cheaper per token; Vera Rubin NVL72 delivers 9.4× the tok/s/chip. (Numbers reflect the default agentic-traces · fp4 selection for this URL — table and chart below update if you change sequence, precision, or model in the controls.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | MI355X37,441Vera Rubin NVL72491,079 | MI355X24,619Vera Rubin NVL72212,232 | MI355X20,726Vera Rubin NVL72195,094 |
| Cost ($/M tok) | MI355X$0.011Vera Rubin NVL72$0.002 | MI355X$0.017Vera Rubin NVL72$0.005 | MI355X$0.020Vera Rubin NVL72$0.005 |
| tok/s/MW | MI355X17,914,118Vera Rubin NVL72148,811,881 | MI355X11,779,513Vera Rubin NVL7264,312,768 | MI355X9,916,649Vera Rubin NVL7259,119,472 |
| Concurrency | MI355X~7Vera Rubin NVL72~423 | MI355X~4Vera Rubin NVL72~133 | MI355X~2Vera Rubin NVL72~90 |