Starstruk Logo MarkStarstruk

v0.34.4

Mesh-LLM/mesh-llmv0.34.4Mar 14, 2026by michaelneale

AI Summary

This update standardizes VRAM-aware context sizes to 64K and resolves issues with draft models and tool calling.

Key Highlights

  • VRAM-aware context set to 64K for all models
  • 128K context hanging with draft models is fixed
  • Pi tool calling is fixed on all machines

New Features

  • VRAM-aware context size of 64K for all models
  • Pi tool calling fix

Full Release Notes

VRAM-aware context: 64K for all models. 128K hung with draft models. Fixes pi tool calling on all machines.