

| Version | Build | OS | Arch | Last Updated | Download URL |
|---|---|---|---|---|---|
0.4.7 | 1 | Mac | arm64 | 03/07/2026 | Download |
0.4.7 | 1 | Windows | x86_64 | 03/07/2026 | Download |
0.4.7 | 1 | Windows | arm64 | 03/07/2026 | Download |
0.4.7 | 1 | Linux | x86_64 | 03/07/2026 | Download |
0.4.7 | 1 | Linux | x86_64 | 03/07/2026 | Download |
0.4.7 | 1 | Linux | arm64 | 03/07/2026 | Download |
0.4.7 | 1 | Linux | arm64 | 03/07/2026 | Download |
Build 1
reasoning_content and content in API responses" is now ON by default in order to improve compatibility with /v1/chat/completions clients
parallel parameter to /api/v1/load endpointpresence_penalty sampling parameter/v1/responses endpoint erroring on none and xhigh reasoning effort/v1/responses responses included logProbs for MLX models even if message.output_text.logprobs was omitted