v0.3.0

   ## Overview                                                                                                                                                                                                                                                                             
                                                                                                                                                                                                                                                                                           
   llama.cpp 0.3.0 introduces the dots3-note multimodal model (with a new DSA-ISWA KV cache), MTP support for GLM-4.5-Air, and tensor-split (`-sm tensor`) plus multi-sequence rollback fixes for DeepSeek 4. ggml is bumped to v0.22.0 (meta-backend tensor split, per-op Metal kernels with parallel compilation, non-in-place `ggml_clamp`), while mtmd gains dots3-note vision/audio, WebP decoding and a Pillow-accurate resize. The server adds a `LLAMA_SERVER_SLOTS_N_DIFF` debug knob, and the web UI gets tabbed chat navigation.                                        
                                                                                                                                                                                                                                                                                           
   ### New models                                                                                                                                                                                                                                                                          
                                                                                                                                                                                                                                                                                           
   - Add dots3-note model with a new DSA-ISWA KV cache type ([#27060](https://github.com/ggml-org/llama.cpp/pull/27060))                                                                                                                                                                   
                                                                                                                                                                                                                                                                                           
   ### Core changes                                                                                                                                                                                                                                                                        
                                                                                                                                                                                                                                                                                           
   - DeepSeek 4: add tensor-split mode via `-sm tensor` ([#26490](https://github.com/ggml-org/llama.cpp/pull/26490))                                                                                                                                                                       
   - DeepSeek 4: fix rollback with multiple sequences ([#26756](https://github.com/ggml-org/llama.cpp/pull/26756))                                                                                                                                                                         
   - Fix meta tensor split state propagation for tensor parallel ([#27574](https://github.com/ggml-org/llama.cpp/pull/27574))                                                                                                                                                              
   - GLM-4.5-Air: add MTP (multi-token prediction) support ([#26534](https://github.com/ggml-org/llama.cpp/pull/26534))                                                                                                                                                                    
   - bailingmoe3: support DSpark ([#27508](https://github.com/ggml-org/llama.cpp/pull/27508))                                                                                                                                                                                              
   - mamba2: flatten in/out projections to dispatch GEMM instead of GEMV ([#27513](https://github.com/ggml-org/llama.cpp/pull/27513))                                                                                                                                                      
   - Models: use `ggml_rope_set_offset` in deepseek2/4, dflash, minicpm3 and plm ([#27382](https://github.com/ggml-org/llama.cpp/pull/27382))                                                                                                                                              
   - Grammar: parse `\-` in char classes as a literal hyphen ([#27591](https://github.com/ggml-org/llama.cpp/pull/27591))                                                                                                                                                                  
   - Common: add `json.h` abstraction ([#27511](https://github.com/ggml-org/llama.cpp/pull/27511)) with a clang LTO fix ([#27575](https://github.com/ggml-org/llama.cpp/pull/27575))                                                                                                       
   - Common: `fit` moved out of the server and now takes `n_streams` into account ([#27496](https://github.com/ggml-org/llama.cpp/pull/27496))                                                                                                                                             
   - Common: fix draft-mtp with embeddings ([#27400](https://github.com/ggml-org/llama.cpp/pull/27400))                                                                                                                                                                                    
   - Arg: remove the `-no-cnv` CLI option ([#27542](https://github.com/ggml-org/llama.cpp/pull/27542))                                                                                                                                                                                     
                                                                                                                                                                                                                                                                                           
   ### Multi-modality changes                                                                                                                                                                                                                                                              
                                                                                                                                                                                                                                                                                           
   - Support dots3-note vision and audio ([#27524](https://github.com/ggml-org/llama.cpp/pull/27524))                                                                                                                                                                                      
   - Support WebP images via ffmpeg ([#27520](https://github.com/ggml-org/llama.cpp/pull/27520))                                                                                                                                                                                           
   - Fix loading videos with the moov atom at the end of the file ([#27596](https://github.com/ggml-org/llama.cpp/pull/27596))                                                                                                                                                             
   - Use a Pillow-accurate resize algorithm and correct `resize_algo` for all models ([#27594](https://github.com/ggml-org/llama.cpp/pull/27594))                                                                                                                                          
   - Use `ggml_rope_set_offset` in the CLIP graph ([#27521](https://github.com/ggml-org/llama.cpp/pull/27521))                                                                                                                                                                             
                                                                                                                                                                                                                                                                                           
   ### Server changes                                                                                                                                                                                                                                                                      
                                                                                                                                                                                                                                                                                           
   - Add `LLAMA_SERVER_SLOTS_N_DIFF` env var to widen the slot debug diff window ([#27600](https://github.com/ggml-org/llama.cpp/pull/27600))                                                                                                                                              
   - Slot fitting logic moved to the common `fit`, now accounting for `n_streams` ([#27496](https://github.com/ggml-org/llama.cpp/pull/27496))                                                                                                                                             
   - Adopt the common `json.h` abstraction ([#27511](https://github.com/ggml-org/llama.cpp/pull/27511))                                                                                                                                                                                    
                                                                                                                                                                                                                                                                                           
   ### UI changes                                                                                                                                                                                                                                                                          
                                                                                                                                                                                                                                                                                           
   - Tabbed navigation for chat conversations ([#27263](https://github.com/ggml-org/llama.cpp/pull/27263))                                                                                                                                                                                 
   - Fix keyboard shortcuts for the chat tabs navigation ([#27609](https://github.com/ggml-org/llama.cpp/pull/27609))                                                                                                                                                                      
                                                                                                                                                                                                                                                                                           
   ### ggml changes                                                                                                                                                                                                                                                                        
                                                                                                                                                                                                                                                                                           
   - ggml bumped to [v0.22.0](https://github.com/ggml-org/ggml/releases/tag/v0.22.0) (ggml/1607):                                                                                                                                                                                          
     - This release adds tensor-split support to the multi-backend (meta) backend with improved split-state propagation, reworks the Metal kernels into per-op sources with parallel compilation, and fixes `ggml_clamp` to be a proper non-in-place op. It also brings new ops            
 (`POOL_1D`, `PAD_REFLECT_1D`), Q2_K SYCL kernels, MoE bias fusion on OpenCL, and assorted fixes across the CUDA, Metal, SYCL, Vulkan, OpenCL and WebGPU backends. 

## Assets

**Nightly build:** [b10621](https://github.com/ggml-org/llama.cpp/releases/tag/b10621)

## More info

- [Releases and versioning of `ggml-org` projects](https://github.com/ggml-org/ggml/discussions/1579)
- [Help tune Metal Flash-attention kernels](https://github.com/ggml-org/llama.cpp/discussions/27668)

## Changelog since v0.2.0

c1d0e7a00 llama.cpp : bump version to 0.3.0 (#27696)
81191affa sync : ggml
93882361f ggml : bump version to 0.22.0 (ggml/1607)
eb25b7263 grammar : parse \- in char classes as literal hyphen (#27591)
814d84bc9 sycl : mark tq2_0 as not supported (#27660)
5ea87ddad webgpu : fix handling of infinity values during ARGSORT and TOP_K (#27538)
f280b2698 metal : per-device tuned (Q, NE) for flash-attn vec (#26570)
b615f5b4b metal: per-op source split + parallel compile (#26561)
b3c3b96a1 misc : read repetition_penalty from generation_config.json (#27659)
758443071 tests : disable DOTS3NOTE arch test for WebGPU (#27654)
71cc86fa4 convert: fix GLM regression in index_tensors (#27655)
a14dba686 ggml : shorten virtual device naming in CUDA and Metal (#27608)
c1c766da5 webgpu : reorder includes since V that appears in common_decls.tmpl may be defined as K in flash_attn_decls.tmpl if KV_OVERLAP (#27545)
160c6b0bd mtmd: video: fix moov atom at the end of file (#27596)
985b14912 ci : apply ccache-clear with older/min/dry-run to all ccache jobs (#27602)
6036c635e ggml : fix ggml_clamp (#27644)
a130532ae mamba2 : Flatten in/out projections to dispatch GEMM instead of GEMV (#27513)
bf0a29cc1 Deepseek 4: `-sm tensor` (#26490)
c060ca974 model : support MTP in GLM-4.5-Air (#26534)
ccc8fd2ba readme : update links (#27617)
d05f89562 fix: Change chat tabs nav shortcuts (#27609)
8d9af2563 test : fix multi-GPU server tests (#27614)
4a08fa297 test: move tools/parser to tests (#27548)
56db501e7 mtmd: use pillow-accurate algo, correct resize_algo for all models (#27594)
95b8e33e1 ci : add test-llama-archs tensor split for Metal (#27598)
a278dcef0 contrib : recommend waiting for CI before merging (#27603)
e8eed4525 server : add LLAMA_SERVER_SLOTS_N_DIFF (#27600)
ba8e0eddf common : skip device_info loop if it's not going to be printed (#26692)
b0539c43e DeepseekV4: fix rollback with multi-seq (#26756)
d3371929b [Tensor parallel] Fix meta tensor split state propagation (#27574)
8144f3192 ui: Chat Conversation Tabbed navigation (#27263)
6657ded4f vendor : update subprocess.h (#27409)
29ea9412a cuda : add POOL_1D support (#27573)
70adb1b4c common: json.h: fix clang lto (#27575)
3f545becc vulkan : added the PAD_REFLECT_1D operation (#26586)
b21e4de74 mtmd: use ggml_rope_set_offset (#27521)
d9f918d2d common: add json.h abstraction (#27511)
2fb989b9e fit: also take into account n_streams (#27496)
9fee29e94 arg: remove -no-cnv from cli [no ci] (#27542)
e85caa81e ci : Restore ROCm job for Ubuntu (#27399)
2115b73d8 model : support DSpark for bailingmoe3 (#27508)
54ee5ee64 mtmd: support dots3-note vision+audio (#27524)
3a653fea9 ci : add older, min and dry-run options to ccache-clear (#27504)
369e1cd61 ggml: optimize concat op by replacing per-element memcpy with row-level memcpy (#24575)
2c6b141ef common : fix draft-mtp with embeddings (#26352, #27299) (#27400)
867229003 sycl : add Q2_K reordered MMVQ and ESIMD kernels (again) (#27490)
3aeb92462 readme : fix server badge alt (#27533)
2100e5926 readme : update badges (#27531)
d775b8967 mtmd: support webp via ffmpeg (#27520)
3af988fab opencl: fold the gpt-oss MoE per-expert bias adds into the epilogue (op/kernel fusion) (#26431)
9a286ac98 docs: improve Windows build instructions (#27381)
a3b9c23ea ci : fix empty release_id in make-release upload step (#27516)
5a32f7b66 model: add dots3-note (#27060)
873e5d8e3 model: use ggml_rope_set_offset() (#27382)
d7fa69b7d ci : run ccache-clear as the last step of release jobs (#27503)