GB300 NVL72 FP8: MTP vs Off Speculative Decoding
Speculative decoding comparison of MTP versus Off on GB300 NVL72 FP8 (NVIDIA Blackwell) running GLM 5/5.1. Throughput, cost, and interactivity differences across LLM workloads. Use the chart controls below to switch sequences and metrics — same interactions as the main inference chart.
MTP acceptance-rate implementations differ across inference engines. Points from different engines are not directly comparable on the same curve — throughput and cost at matched interactivity may reflect engine-level differences rather than pure speculative decoding gains. Interpret cross-engine comparisons with caution.
MTP posts 6418 tok/s/GPU for $0.12 per million tokens at 42 tok/s/user on GLM 5/5.1 (GB300 NVL72 FP8); Off posts 3237 tok/s/GPU for $0.24. MTP is 112% cheaper per token; MTP delivers 98% more tok/s/GPU. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level.
Throughput at 50 tok/s/user on GLM 5/5.1 (GB300 NVL72 FP8): MTP hits 5729 tok/s/GPU, Off hits 510. Per-million costs land at $0.13 and $1.44 respectively. MTP is 1020% cheaper per token; MTP delivers 1024% more tok/s/GPU. Speculative decoding trades extra compute on draft tokens for fewer decoding steps — the payoff depends on sequence length and batch size.
Toward the upper edge of the 35–65 tok/s/user interactivity band, at 57 tok/s/user on GLM 5/5.1 (GB300 NVL72 FP8): MTP runs 3576 tok/s/GPU at $0.21/M tokens, Off runs 233 at $3.19/M. MTP is 1409% cheaper per token; MTP delivers 1434% more tok/s/GPU. Gains from speculative decoding vary by workload; short-output prompts tend to benefit less. (Numbers reflect this URL's pinned 8k/1k · fp8 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MTP:6417.9Off:3237.2 | MTP:5729.4Off:509.9 | MTP:3576.0Off:233.1 |
| Cost ($/M tok) | MTP:$0.115Off:$0.244 | MTP:$0.129Off:$1.441 | MTP:$0.211Off:$3.192 |
| tok/s/MW | MTP:3056144Off:1541547 | MTP:2728275Off:242821 | MTP:1702863Off:111001 |
| Concurrency | MTP:~2312Off:~799 | MTP:~1721Off:~132 | MTP:~867Off:~36 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.