B300 FP4: M3 EAGLE vs Off Speculative Decoding
Speculative decoding comparison of M3 EAGLE versus Off on B300 FP4 (NVIDIA Blackwell) running MiniMax M3 428B. 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 19–237 tok/s/user interactivity band, at 73 tok/s/user on MiniMax M3 428B (B300 FP4): M3 EAGLE runs 13730 tok/s/GPU at $0.05/M tokens, Off runs 5941 at $0.11/M. M3 EAGLE is 131% cheaper per token; M3 EAGLE delivers 131% more tok/s/GPU. Gains from speculative decoding vary by workload; short-output prompts tend to benefit less.
At 128 tok/s/user on MiniMax M3 428B (B300 FP4), M3 EAGLE delivers 7414 tok/s/GPU at $0.09 per million tokens; Off delivers 2866 tok/s/GPU at $0.23. M3 EAGLE is 162% cheaper per token; M3 EAGLE delivers 159% more tok/s/GPU. Speculative decoding accepts draft tokens to reduce per-token latency — gains vary by workload and prompt distribution.
M3 EAGLE posts 5701 tok/s/GPU for $0.11 per million tokens at 182 tok/s/user on MiniMax M3 428B (B300 FP4); Off posts 1214 tok/s/GPU for $0.54. M3 EAGLE is 370% cheaper per token; M3 EAGLE delivers 369% more tok/s/GPU. Draft-token acceptance rates determine whether speculative decoding helps or hurts at a given concurrency level. (Numbers reflect this URL's pinned 8k/1k · fp4 workload — changing sequence or model updates both the table and chart; the table stays pinned to this page's precision, so precision toggles in the controls affect the chart only.)
| Metric | Interactivity (tok/s/user) | Interactivity (tok/s/user) | Interactivity (tok/s/user) |
|---|---|---|---|
| Throughput (tok/s/gpu) | M3 EAGLE:13730.3Off:5940.5 | M3 EAGLE:7414.1Off:2866.1 | M3 EAGLE:5700.5Off:1214.2 |
| Cost ($/M tok) | M3 EAGLE:$0.047Off:$0.110 | M3 EAGLE:$0.087Off:$0.229 | M3 EAGLE:$0.114Off:$0.537 |
| tok/s/MW | M3 EAGLE:7226475Off:3126584 | M3 EAGLE:3902136Off:1508460 | M3 EAGLE:3000277Off:639049 |
| Concurrency | M3 EAGLE:~503Off:~430 | M3 EAGLE:~64Off:~12 | M3 EAGLE:~64Off:~8 |
Inference Performance
Inference performance metrics across different models, hardware configurations, and serving parameters.