GB200 NVL72 FP4: MTP vs Off Speculative Decoding
Speculative decoding comparison of MTP versus Off on GB200 NVL72 FP4 (NVIDIA Blackwell) running DeepSeek R1. 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.
Near the low end of the 19–175 tok/s/user interactivity band, at 57 tok/s/user on DeepSeek R1 (GB200 NVL72 FP4): MTP runs 11333 tok/s/GPU at $0.05/M tokens, Off runs 7173 at $0.08/M. MTP is 57% cheaper per token; MTP delivers 58% more tok/s/GPU. Gains from speculative decoding vary by workload; short-output prompts tend to benefit less.
At 97 tok/s/user on DeepSeek R1 (GB200 NVL72 FP4), MTP delivers 7092 tok/s/GPU at $0.09 per million tokens; Off delivers 2954 tok/s/GPU at $0.20. MTP is 123% cheaper per token; MTP delivers 140% more tok/s/GPU. Speculative decoding accepts draft tokens to reduce per-token latency — gains vary by workload and prompt distribution.
MTP posts 3059 tok/s/GPU for $0.26 per million tokens at 136 tok/s/user on DeepSeek R1 (GB200 NVL72 FP4); Off posts 320 tok/s/GPU for $2.46. MTP is 847% cheaper per token; MTP delivers 856% more tok/s/GPU. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level. (Numbers reflect this URL's pinned 1k/1k · fp4 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MTP:11333.4Off:7172.9 | MTP:7092.5Off:2953.6 | MTP:3058.6Off:320.1 |
| Cost ($/M tok) | MTP:$0.054Off:$0.084 | MTP:$0.089Off:$0.200 | MTP:$0.260Off:$2.459 |
| tok/s/MW | MTP:5396835Off:3415660 | MTP:3377367Off:1406496 | MTP:1456481Off:152420 |
| Concurrency | MTP:~3397Off:~2177 | MTP:~2028Off:~512 | MTP:~686Off:~35 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.