Qwen 3.5 397B-A17B · B200 · Precision comparison
B200: FP4 vs BF16 precision comparison
How FP4 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.
FP4 posts 4,731 tok/s/chip for $0.10 per million tokens at 112 tok/s/user on Qwen 3.5 397B-A17B (B200); BF16 posts 1,216 tok/s/chip for $0.40. FP4 delivers 3.9× the tok/s/chip on the same chip, which makes it 74% cheaper per token. Quantization-level accuracy differences are tracked on the evaluation tab.
Throughput at 166 tok/s/user on Qwen 3.5 397B-A17B (B200): FP4 hits 1,864 tok/s/chip, BF16 hits 660. Costs per million tokens land at $0.26 and $0.73 respectively. FP4 delivers 182% more tok/s/chip on the same chip, which makes it 65% 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 59–272 tok/s/user interactivity band, at 219 tok/s/user on Qwen 3.5 397B-A17B (B200): FP4 runs 1,014 tok/s/chip at $0.47 per million tokens, BF16 runs 299 at $1.60. FP4 delivers 3.4× the tok/s/chip on the same chip, which makes it 70% cheaper per token. Precision changes affect both inference speed and model quality — consult the evaluation tab for accuracy benchmarks. (Numbers reflect the default 1k/1k selection for this URL — table and chart below update if you change sequence or model in the controls. Each side uses the best available serving configuration for that precision, which may include speculative decoding such as MTP where recipes exist — the same convention as the other comparison pages.)
Interpolated from real benchmark data. Edit target interactivity values below to compare at different operating points.
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/chip) | FP44,731BF161,216 | FP41,864BF16660 | FP41,014BF16299 |
| Cost ($/M tok) | FP4$0.10BF16$0.40 | FP4$0.26BF16$0.73 | FP4$0.47BF16$1.60 |
| tok/s/MW | FP42,766,489BF16711,170 | FP41,090,275BF16386,094 | FP4592,803BF16175,125 |
| Concurrency | FP4~199BF16~49 | FP4~51BF16~18 | FP4~5BF16~6 |
