What's Changed
- lemon.cpp by @jeremyfowers in #487
- Fix all known C++ tray, installer, and cli bugs (remake) by @jeremyfowers in #496
- Add enable_thinking parameter for Qwen3 GGUF models by @Kritik-07 in #490
- C++: Implement delete endpoint for flm by @jeremyfowers in #493
- Memory tracking with llamacpp by @amd-pworfolk in #483
- Fix rocm support in C++ by @jeremyfowers in #498
- C++: log streaming to web app by @jeremyfowers in #503
- C++ Linux Support and .deb installer by @jeremyfowers in #497
- Fix install options on mac/linux by @jeremyfowers in #506
- C++: Overhaul the model manager by @jeremyfowers in #509
- C++: Fix max_completion_tokens in rai-serve by @jeremyfowers in #513
- Fix: Remove redundant callback from webui by @jeremyfowers in #512
- Improve FLM update policy by @jeremyfowers in #505
- C++: Fix port assignment bug by @jeremyfowers in #514
- C++: support halt and reasoning in ryzenai-serve by @jeremyfowers in #516
- Fix bug in performance table row merging by @amd-pworfolk in #511
- C++: Polish test workflow by @jeremyfowers in #518
- C++: disable self-hosted ubuntu jobs by @jeremyfowers in #522
- C++: Add context size to the health endpoint by @jeremyfowers in #519
- Overhaul C++ versioning by @jeremyfowers in #520
- C++: rename ryzenai-serve to ryzenai-server by @jeremyfowers in #521
- C++: Improved stop command by @jeremyfowers in #524
- C++: Polish the CLI by @jeremyfowers in #526
- Increase default context and batch sizes for embedding models by @ramkrishna2910 in #510
- lm-eval fix by @ramkrishna2910 in #417
- C++: Polish ryzenai-server by @jeremyfowers in #527
- C++: Fix quit and ctrl+c by @jeremyfowers in #528
- Release C++ open beta and python 8.2.1 by @jeremyfowers in #529
Full Changelog: v8.2.0...v8.2.1