Browsing: multi-token prediction MTP

If you’re running LLM inference at scale and haven’t looked at multi-token prediction MTP yet, you’re leaving real latency gains…