🚀 Foundry Local v2.3.0
We're excited to announce Foundry Local v2.3.0. This release improves reliability and inference across SDKs and EPs.
🆕 What's New
⚡Improvements
- Better support for coding models
- Better inference performance across EPs
- Add language support to Java PCM and WAV transcription
- Add a native shared model cache across Foundry Local SDKs
🐛 Fixed
- Register CUDA EP before TRTRTX EP, fixing occasional OOM errors
- Apply chat defaults correctly
- Fixed support for Parakeet and Whisper models
- Fix hang preventing shutdown
Note
CUDA and cuDNN runtimes must be installed separately; Foundry Local does not download them.
- Windows/Linux x64: CUDA 12 and compatible cuDNN 9 runtimes.
- Windows/Linux ARM64: CUDA 13 and compatible cuDNN 9 runtimes.
On Windows, add the runtime DLL directories to PATH. On Linux, configure the system library search path or LD_LIBRARY_PATH.
Configure library search paths before starting your application.
Windows x64 applications must additionally call AddDllDirectory for the CUDA runtime directory before CUDA execution-provider registration; PATH alone is insufficient for the GenAI CUDA bridge. This additional step is not required for the shipped Windows ARM64 bridge.
📥 Source Code
GitHub automatically includes source archives with the v2.3.0 release:
📚 Resources
| 📖 Documentation | https://aka.ms/foundry-local-docs |
| 🐙 GitHub | https://github.com/microsoft/Foundry-Local |
| 🧪 Samples | https://github.com/microsoft/Foundry-Local/tree/main/samples |
| 💬 Discord | https://aka.ms/foundry-local-discord |
💙 Feedback
File issues at https://github.com/microsoft/Foundry-Local/issues and include the SDK version, platform, and relevant diagnostics.