llama-router, the C++ "llama-swap" for llama.cpp#17629

Draft

ServeurpersoCom wants to merge 29 commits intoggml-org:masterfrom

ServeurpersoCom:llama-router

Commits on Nov 30, 2025

llama-router: multi-model serving with dynamic backends
ServeurpersoCom
committed
llama-router: fix logging init via static constructor
ServeurpersoCom
committed
llama-router: centralize defaults and add mmproj auto-detection
ServeurpersoCom
committed
llama-router: add process grouping for selective VRAM management
ServeurpersoCom
committed
llama-router: add legacy endpoint support for single-model compat
ServeurpersoCom
committed
llama-router: add comprehensive debug logging
ServeurpersoCom
committed
llama-router: implement SSE streaming and production safety features
ServeurpersoCom
committed
llama-router: fix segfault from static initialization order fiasco
ServeurpersoCom
committed
llama-router: auto-detect sibling binary, capture logs, wait for backend ready
ServeurpersoCom
committed
llama-router: implement cross-platform subprocess I/O forwarding and lifecycle monitoring
ServeurpersoCom
committed
llama-router: validate binary before spawn, clean child error handling
ServeurpersoCom
committed
llama-router: add multi-engine support with configurable spawn and endpoints
ServeurpersoCom
committed
llama-router: fix SSE streaming termination and use-after-free
ServeurpersoCom
committed
llama-router: auto-rescan, admin endpoints, and fixes
ServeurpersoCom
committed
llama-router: add --import-dir for custom model collections
ServeurpersoCom
committed
llama-router: add README with CLI reference and configuration guide
ServeurpersoCom
committed
llama-router: document KISS philosophy, optimization patterns, and system architecture
ServeurpersoCom
committed
llama-router: fix PATH binary support and macOS detection
ServeurpersoCom
committed
llama-router: separate quick-start guide from technical architecture docs
ServeurpersoCom
committed
llama-router: async polling for process termination after SIGKILL
ServeurpersoCom
committed
llama-router: separate PROCESS (OS) and BACKEND (HTTP) polling constants
ServeurpersoCom
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

llama-router, the C++ "llama-swap" for llama.cpp#17629

llama-router, the C++ "llama-swap" for llama.cpp#17629
ServeurpersoCom wants to merge 29 commits intoggml-org:masterfrom
ServeurpersoCom:llama-router

Commits on Nov 30, 2025

llama-router: multi-model serving with dynamic backends

llama-router: fix logging init via static constructor

llama-router: centralize defaults and add mmproj auto-detection

llama-router: add process grouping for selective VRAM management

llama-router: add legacy endpoint support for single-model compat

llama-router: add comprehensive debug logging

llama-router: implement SSE streaming and production safety features

llama-router: fix segfault from static initialization order fiasco

llama-router: auto-detect sibling binary, capture logs, wait for backend ready

llama-router: implement cross-platform subprocess I/O forwarding and lifecycle monitoring

llama-router: validate binary before spawn, clean child error handling

llama-router: add multi-engine support with configurable spawn and endpoints

llama-router: fix SSE streaming termination and use-after-free

llama-router: auto-rescan, admin endpoints, and fixes

llama-router: add --import-dir for custom model collections

llama-router: add README with CLI reference and configuration guide

llama-router: document KISS philosophy, optimization patterns, and system architecture

llama-router: fix PATH binary support and macOS detection

llama-router: separate quick-start guide from technical architecture docs

llama-router: async polling for process termination after SIGKILL

llama-router: separate PROCESS (OS) and BACKEND (HTTP) polling constants

Commits on Dec 1, 2025

llama-router: add real-time model swap notifications via SSE

llama-router: document notify_model_swap feature in README and ARCHITECTURE

llama-router: add embedded WebUI support

llama-router: add startup_model configuration option

llama-router: document startup_model in README and ARCHITECTURE

llama-router: auto-configure startup_model on first HF download

llama-router: add --jinja to default spawn configuration

llama-router: replace implicit arg injection with explicit placeholders

llama-router, the C++ "llama-swap" for llama.cpp#17629ServeurpersoCom wants to merge 29 commits intoggml-org:masterggml-org/llama.cpp:masterfrom ServeurpersoCom:llama-routerServeurpersoCom/llama.cpp:llama-routerCopy head branch name to clipboard

Commits

Commits on Nov 30, 2025

Commits on Dec 1, 2025

llama-router, the C++ "llama-swap" for llama.cpp#17629
ServeurpersoCom wants to merge 29 commits intoggml-org:masterfrom
ServeurpersoCom:llama-router