Disable MTP when coding: Tests show speed drops drastically at long context
fbms2 · reddit · 2026-08-22
Tests indicate that Multi-Token Prediction (MTP) is counterproductive for coding. While MTP achieves 140t/s in chat mode on a 5090, coding speeds crash from 100t/s to 10-20t/s once the context exceeds 60k. Disabling MTP results in a stable 40t/s even at very high context lengths, despite a lower starting speed.
More from coding & agent
- Veracode: AI now writes ~half of committed code, but security pass rate stuck at 56% — WeldPond · 2026-10-16
- Engineer: my job went from silently thinking at a computer to constantly talking to it — KevinNaughtonJr · 2026-10-16
- LangChain launches Managed Deep Agents: build production agents as a directory of files — hwchase17 · 2026-10-16
- Vercel Goes Full Backend: Python GA, Containers, and Daemons — cramforce · 2026-10-16
- WorldMemArena Accepted at NeurIPS: Memory Storage ≠ Memory Use — xwang_lk · 2026-10-16
- Vercel Ship Demos Container Deploys for Go, Ruby, Java and More — edwin · 2026-10-16