H200 FP8: MTP vs Off Speculative Decoding
Speculative decoding comparison of MTP versus Off on H200 FP8 (NVIDIA Hopper) running DeepSeek V4 Pro 1.6T. 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.
At 26 tok/s/user on DeepSeek V4 Pro 1.6T (H200 FP8), MTP delivers 273 tok/s/GPU at $1.43 per million tokens; Off delivers 299 tok/s/GPU at $1.32. Off is 8% cheaper per token; Off delivers 10% more tok/s/GPU. Speculative decoding accepts draft tokens to reduce per-token latency — gains vary by workload and prompt distribution.
MTP posts 220 tok/s/GPU for $1.80 per million tokens at 44 tok/s/user on DeepSeek V4 Pro 1.6T (H200 FP8); Off posts 158 tok/s/GPU for $2.49. MTP is 38% cheaper per token; MTP delivers 39% more tok/s/GPU. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level.
Throughput at 62 tok/s/user on DeepSeek V4 Pro 1.6T (H200 FP8): MTP hits 163 tok/s/GPU, Off hits 77. Per-million costs land at $2.34 and $4.93 respectively. MTP is 111% cheaper per token; MTP delivers 110% more tok/s/GPU. Speculative decoding trades extra compute on draft tokens for fewer decoding steps — the payoff depends on sequence length and batch size. (Numbers reflect this URL's pinned 1k/1k · fp8 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MTP:272.7Off:299.3 | MTP:219.5Off:157.6 | MTP:162.9Off:77.4 |
| Cost ($/M tok) | MTP:$1.433Off:$1.324 | MTP:$1.802Off:$2.486 | MTP:$2.342Off:$4.931 |
| tok/s/MW | MTP:157624Off:173019 | MTP:126897Off:91093 | MTP:94160Off:44762 |
| Concurrency | MTP:~54Off:~53 | MTP:~21Off:~16 | MTP:~11Off:~5 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.