Tagged "multimodal-embeddings"

3 articles tagged multimodal-embeddings, 7 October 2026 to 8 October 2026. Newest first.

• Ollama v0.40.0 Brings MLX Runtime to Apple Silicon and Decision Models 8 October 2026

  Ollama's latest release automatically routes supported model architectures to the MLX runtime on Apple Silicon devices, improving inference performance. The release also adds support for decision models and multimodal embeddings including EmbeddingGemma2.

• Ollama v0.40.0 Brings Multimodal Embeddings and MLX Default on Apple Silicon 7 October 2026

  Ollama's latest release makes MLX the default runtime for supported models on Apple Silicon and implements native multimodal embedding support through EmbeddingGemma2Model architecture. This brings efficient text, audio, and video embeddings to on-device deployment with significantly reduced memory footprint.

• Google DeepMind Releases EmbeddingGemma 2: A 740M Multimodal Embedding Model for On-Device AI 7 October 2026

  Google DeepMind has released EmbeddingGemma 2, an open-source multimodal embedding model that processes text, audio, and video while running in just 567MB of memory. Built on Gemma 4, this 740M parameter model enables efficient semantic search and retrieval on-device without cloud dependencies.