Coding Tools
llama.cpp exposes reasoning effort levels in server
llama.cpp Releaseshigh confidence
Why it matters
- →Developers can now programmatically control how deeply an AI model thinks before responding.
- →Clients can probe for supported reasoning levels to optimize latency versus accuracy trade-offs.
- →Standardizes the interface for extended thinking models in local inference servers.
llama.cpp has updated its server component to expose reasoning effort levels within the chat template interface. The update allows clients to probe for supported levels—ranging from minimal to max—and serialize these capabilities as JSON properties. This change enables developers to explicitly control the depth of a model's internal reasoning process during inference. The feature is available across existing builds including CUDA, ROCm, and Apple Silicon variants.
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 Update Fixes Reasoning Budget Handling
llama.cpp Releases · July 13, 2026 · Same story
- 2Fine-tunes reduce Qwen3.8-27B reasoning tokens
Sam Witteveen · October 4, 2026 · Related
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
