Skip to content

Releases: ceveyne/qwen3-vl-embedding

v2026.07.26-b9951

Choose a tag to compare

@github-actions github-actions released this 26 Jul 06:29

Changed

  • Release packaging now rejects unsafe runtime library links.

The included binaries are ready to use on macOS with Apple Silicon (arm64). Windows, Linux, and Intel Macs are not supported by this release.

v2026.07.20-b9951

Choose a tag to compare

@github-actions github-actions released this 20 Jul 05:18

Added

  • Local reranking can now use task-specific instructions for more precise visual, text, and multimodal results.

The included binaries are ready to use on macOS with Apple Silicon (arm64). Windows, Linux, and Intel Macs are not supported by this release.

v2026.07.16-b9951

Choose a tag to compare

@github-actions github-actions released this 16 Jul 19:18

Fixed

  • Prevented llama-server crashes when long embedding or reranking prompts are processed across multiple 512-token decode chunks.

The included binaries are ready to use on macOS with Apple Silicon (arm64). Windows, Linux, and Intel Macs are not supported by this release.

v2026.07.15-b9951

Choose a tag to compare

@github-actions github-actions released this 15 Jul 12:26
  • Added Qwen3-VL reranking support alongside embeddings, making it easier to rank image, text, and mixed-content results locally.

The included binaries are ready to use on macOS with Apple Silicon (arm64). Windows, Linux, and Intel Macs are not supported by this release.

v2026.07.11-b9951

Choose a tag to compare

@github-actions github-actions released this 11 Jul 12:20
  • Added prebuilt binaries for Qwen3-VL embeddings, so you can get started without building from source.

The included binaries are ready to use on macOS with Apple Silicon (arm64). Windows, Linux, and Intel Macs are not supported by this release.