Skip to content

v20260106rc0

Choose a tag to compare

@github-actions github-actions released this 06 Jan 18:23
· 66 commits to main since this release

nomic-embed-v1.5

Multimodal embedding server supporting both text and image embeddings.

Installation

Download the tarball for your architecture and model variant:

# x86_64 - Quantized (smaller, faster, ~99% quality)
tar -xzf nomic-embed-v1.5-20260106rc0-linux-x64-quantized.tar.gz

# x86_64 - Full precision (larger, best quality)
tar -xzf nomic-embed-v1.5-20260106rc0-linux-x64-full.tar.gz

# ARM64 - Quantized (Raspberry Pi, AWS Graviton, Apple Silicon Linux VMs)
tar -xzf nomic-embed-v1.5-20260106rc0-linux-arm64-quantized.tar.gz

# ARM64 - Full precision
tar -xzf nomic-embed-v1.5-20260106rc0-linux-arm64-full.tar.gz

# macOS (Apple Silicon M1/M2/M3) - Quantized
tar -xzf nomic-embed-v1.5-20260106rc0-macos-arm64-quantized.tar.gz

# macOS (Apple Silicon M1/M2/M3) - Full precision
tar -xzf nomic-embed-v1.5-20260106rc0-macos-arm64-full.tar.gz

Running

cd nomic-embed-v1.5-*
./nomic-serve
# Server starts on http://localhost:8080

API Endpoints

  • POST /txt/embed - Single text embedding
  • POST /txt/batch - Batch text embeddings
  • POST /txt/query - Query with enforced search_query prefix
  • POST /img/embed - Single image embedding
  • POST /img/batch - Batch image embeddings
  • GET /health - Health check
  • GET /docs - Swagger UI

Environment Variables

  • PORT - Server port (default: 8080)
  • TXT_MODEL - Path to text ONNX model (default depends on variant: model_quantized.onnx or model.onnx)
  • TOKENIZER - Path to tokenizer file (default: models/txt/tokenizer.json)
  • IMG_MODEL - Path to vision ONNX model (default depends on variant: model_quantized.onnx or model.onnx)
  • USE_GPU - Set to 1 to enable GPU (requires CUDA drivers)

Model Variants:

  • Quantized (*-quantized.tar.gz): Smaller files, faster inference, ~99% quality. Recommended for most use cases.
  • Full (*-full.tar.gz): Larger files, best quality (fp32 precision). Use for maximum accuracy.

Note: This release includes CPU-only runtime. For GPU support, use the Docker images (mindthemath/nomic-embed-v1.5-rs:*-gpu) or manually install CUDA libraries.

Full Changelog: https://github.com/mindthemath/nomic-api-rs/commits/v20260106rc0