feat(dllm): DiffusionGemma backend via dllm.cpp with native gemma4 chat - #10258

Closed

localai-bot wants to merge 12 commits into

feat/dllm-backend

Commits on Jul 29, 2026

feat(dllm): purego backend scaffold over the dllm.cpp C-ABI

mudler
authored and
localai-org-maint-bot
committed
feat(dllm): gemma4 streaming parser emitting ChatDeltas

mudler
authored and
localai-org-maint-bot
committed
feat(dllm): rich gRPC backend with ChatDelta streaming

mudler
authored and
localai-org-maint-bot
committed
feat(dllm): backend packaging, gallery index, CI matrix

mudler
authored and
localai-org-maint-bot
committed
feat(dllm): diffusiongemma gallery entry and e2e coverage

mudler
authored and
localai-org-maint-bot
committed
docs(dllm): backend documentation and agents topic guide

mudler
authored and
localai-org-maint-bot
committed
chore(dllm): review fixes - file modes and build-matrix doc accuracy

mudler
authored and
localai-org-maint-bot
committed
feat(grpc): request cancellation for Go backends via the Cancellable capability

mudler
authored and
localai-org-maint-bot
committed
docs(dllm): record Q4_K_M validation and quantization guidance

mudler
authored and
localai-org-maint-bot
committed
feat(dllm): default gallery entry on Q4_K_M; add Q8_0 variant

mudler
authored and
localai-org-maint-bot
committed
feat(dllm): image input through the backend (multimodal C-ABI)

mudler
authored and
localai-org-maint-bot
committed
chore(dllm): bump dllm.cpp pin to P5 head

mudler
authored and
localai-org-maint-bot
committed