llama.cpp New Command: One-Line Serve with MTP Speculative Decoding

ggerganov · x · 2026-08-15

ggerganov demonstrates the llama serve command in llama.cpp, enabling MTP speculative decoding with --spec-type draft-mtp, plus advanced options like --reasoning-preserve and --agent, simplifying local deployment.

Related event: llama.cpp Adds One-Command MTP Speculative Decoding(2 posts)→

Original post →

More from coding & agent

coding & agent channel →