Coding Tools
llama.cpp adds batched tensor copy API
llama.cpp Releaseshigh confidence
Why it matters
- →Reduces I/O overhead by batching tensor copies instead of processing them individually.
- →Improves performance for model loading and context switching on CUDA and other backends.
- →Optimizes internal data movement without requiring changes to user-facing code.
The llama.cpp project has merged a significant backend update introducing a batched tensor copy API to its ggml library. The change restructures how data is moved between devices, grouping operations by buffer type to reduce overhead during context I/O and model loading. CUDA support is included in this initial rollout, with tests verifying dispatch logic across mock and real backends. This optimization aims to improve efficiency for users running complex models or frequent context switches.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1llama.cpp b10705 Release Enhances Tensor Handling
llama.cpp Releases · August 31, 2026 · Same story
- 2llama.cpp b10984 release enhances CUDA support
llama.cpp Releases · September 16, 2026 · Same story
More from llama.cpp Releases
Coding Toolscoding
llama.cpp b12022 adds ROCm 10 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally bringing ROCm 10.0 to Linux and Windows alongside CUDA 13.4, effectively closing the hardware gap for AMD users who were previously left behind. The inclusion of native Snapdragon support on Linux marks a significant step toward running local LLMs on ARM-based mobile chips without emulation overhead. While KleidiAI on Apple Silicon is disabled in this build, the broader expansion to diverse accelerators ensures that developers aren't locked into NVIDIA ecosystems anymore.
llama.cpp Releases
Coding Toolscoding
llama.cpp b12023 fixes shared prefix bug
This release addresses a specific inference artifact where decision prefixes were incorrectly bleeding into output embeddings. By capping shared prefixes at the first output position, llama.cpp ensures that option tokens remain distinct for parent and child tasks. This is a targeted fix for multi-step reasoning or voting workflows rather than a performance boost. It stabilizes complex generation patterns without changing the underlying model architecture.
More in Coding Tools
Coding Toolscoding
Claude Code v2.1.316: Subagents and Gateway Budgets
Anthropic is tightening control over Claude Code's autonomous behavior with new environment variables that let you disable 'ultrathink' or cap subagent effort levels. For teams, the introduction of pooled group budgets on the gateway allows shared monthly spending limits across members, a practical fix for enterprise cost management. The release also patches critical bugs where shell commands would hang or crash on exit, stabilizing the developer experience significantly.
Claude Code Releases
© The Verge AICoding Toolscoding
