B300 FP4: MTP vs Off Speculative Decoding
Speculative decoding comparison of MTP versus Off on B300 FP4 (NVIDIA Blackwell) 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.
Near the low end of the 10–101 tok/s/user interactivity band, at 33 tok/s/user on DeepSeek V4 Pro 1.6T (B300 FP4): MTP runs 9706 tok/s/GPU at $0.07/M tokens, Off runs 8350 at $0.08/M. MTP is 15% cheaper per token; MTP delivers 16% more tok/s/GPU. Gains from speculative decoding vary by workload; short-output prompts tend to benefit less.
At 56 tok/s/user on DeepSeek V4 Pro 1.6T (B300 FP4), MTP delivers 7882 tok/s/GPU at $0.08 per million tokens; Off delivers 1827 tok/s/GPU at $0.36. MTP is 334% cheaper per token; MTP delivers 331% more tok/s/GPU. Speculative decoding accepts draft tokens to reduce per-token latency — gains vary by workload and prompt distribution.
MTP posts 5096 tok/s/GPU for $0.13 per million tokens at 79 tok/s/user on DeepSeek V4 Pro 1.6T (B300 FP4); Off posts 551 tok/s/GPU for $1.22. MTP is 858% cheaper per token; MTP delivers 825% more tok/s/GPU. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level. (Numbers reflect this URL's pinned 8k/1k · fp4 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | MTP:9706.3Off:8350.5 | MTP:7881.6Off:1827.4 | MTP:5096.1Off:550.9 |
| Cost ($/M tok) | MTP:$0.067Off:$0.077 | MTP:$0.082Off:$0.358 | MTP:$0.127Off:$1.221 |
| tok/s/MW | MTP:5108582Off:4394982 | MTP:4148189Off:961773 | MTP:2682182Off:289958 |
| Concurrency | MTP:~3072Off:~999 | MTP:~927Off:~33 | MTP:~256Off:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.