Qwen 3.5 397B-A17B · B200 · Precision comparison

B200: FP8 vs BF16 precision comparison

How FP8 and BF16 precision affect Qwen 3.5 397B-A17B inference on B200 (NVIDIA Blackwell). Throughput, latency, and cost across LLM workloads. Use the chart controls below to switch sequences and metrics — same interactions as the main inference chart.

FP8 posts 6,809 tok/s/chip for $0.071 per million tokens at 101 tok/s/user on Qwen 3.5 397B-A17B (B200); BF16 posts 3,116 tok/s/chip for $0.15. FP8 delivers 119% more tok/s/chip on the same chip, which makes it 54% cheaper per token. Quantization-level accuracy differences are tracked on the evaluation tab.

Throughput at 157 tok/s/user on Qwen 3.5 397B-A17B (B200): FP8 hits 4,802 tok/s/chip, BF16 hits 2,011. Costs per million tokens land at $0.10 and $0.24 respectively. FP8 delivers 139% more tok/s/chip on the same chip, which makes it 58% cheaper per token. The cost-throughput tradeoff from lower precision is only part of the picture — see the evaluation page for accuracy data.

Toward the upper edge of the 47–267 tok/s/user interactivity band, at 212 tok/s/user on Qwen 3.5 397B-A17B (B200): FP8 runs 3,596 tok/s/chip at $0.13 per million tokens, BF16 runs 1,296 at $0.37. FP8 delivers 178% more tok/s/chip on the same chip, which makes it 64% cheaper per token. Precision changes affect both inference speed and model quality — consult the evaluation tab for accuracy benchmarks. (Numbers reflect the default 8k/1k selection for this URL — table and chart below update if you change sequence or model in the controls. Each side uses the best available serving configuration for that precision, which may include speculative decoding such as MTP where recipes exist — the same convention as the other comparison pages.)

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)
FP86,809BF163,116
FP84,802BF162,011
FP83,596BF161,296
Cost ($/M tok)
FP8$0.071BF16$0.15
FP8$0.10BF16$0.24
FP8$0.13BF16$0.37
tok/s/MW
FP83,981,704BF161,822,092
FP82,808,282BF161,175,989
FP82,103,152BF16757,728
Concurrency
FP8~34BF16~31
FP8~15BF16~13
FP8~8BF16~6
Qwen 3.5 397B-A17B: FP8 versus BF16 throughput and cost at matched interactivity levels on B200
FP8 versus BF16 throughput and cost per million tokens on B200 for this comparison's canonical default workload
Open full-size image ↗