[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f6uiQyIkqaRxZqlaTZLISdXi8IILVPPhs0a9VzG9jKWQ":3},{"article":4,"related":18},{"id":5,"slug":6,"title":7,"seo_title":7,"description":8,"keywords":9,"content":10,"category":11,"image_url":12,"source_guid":13,"published_at":14,"created_at":14,"updated_at":15,"source_url":16,"source_name":17},1347,"google-releases-embeddinggemma-2-for-local-multimodal-search","Google releases EmbeddingGemma 2 for local multimodal search","Google’s EmbeddingGemma 2 embeds text, images, video, audio and code locally. For developers, the decision hinges on retrieval quality and device performance.","[\"EmbeddingGemma 2\",\"Google\",\"multimodal search\",\"local embeddings\",\"offline RAG\"]","\u003Cp>Google has released EmbeddingGemma 2, an open model that turns text, images, video, audio and code into numerical vectors for finding and comparing similar content, \u003Ca href=\"https:\u002F\u002Fthe-decoder.com\u002Fgoogle-claims-embeddinggemma-2-outperforms-rival-embedding-models-twice-its-size\u002F\" rel=\"noopener noreferrer\">according to the-decoder.com\u003C\u002Fa>, citing Google. The model has 740 million parameters. Google claims it beats competing models up to twice its size on multimodal embedding benchmarks.\u003C\u002Fp>\n\u003Cp>The report says the model runs locally without an API key and gives browser query times of about 20 to 70 milliseconds through WebGPU. It also reports roughly 191 MB of RAM usage and local vector database storage reductions of up to six times. These are reported performance figures, not results independently verified here. Weights are available on Hugging Face and Kaggle.\u003C\u002Fp>\n\u003Cp>For developers choosing a search model, the first decision is whether the application needs multiple media types. The report also describes a 270-million-parameter option for text-only tasks. That makes the smaller version a sensible evaluation starting point for a text-only collection. For a mixed collection, test whether the larger model retrieves the right material for representative user queries across the media types the application actually uses.\u003C\u002Fp>\n\u003Cp>The benchmark claim suggests compact models deserve consideration, but it does not settle that choice. Before replacing an existing embedding model, compare relevant results on the same documents and queries. Measure browser latency, memory use and index size on the devices intended to run the application. Treat the reported speed and storage figures as comparisons to investigate, rather than capacity assumptions.\u003C\u002Fp>\n\u003Cp>The report says pairing EmbeddingGemma 2 with small open models such as Gemma 4 can support offline retrieval-augmented generation without sending data to external servers. For teams pursuing that setup, the useful acceptance test is the complete application: disconnect it from the network and check that ingestion, retrieval and answer generation still work. Separately inspect network activity during normal operation before concluding that the deployed application keeps its data local.\u003C\u002Fp>","AI & Machine Learning","https:\u002F\u002Fseedwire.co\u002Fapi\u002Fimages\u002Farticles\u002F1791354143635-sqvlc5galw.webp","5a5fb1b4aeb3d1d34e075c720da5f5391d1ebc7272bfd34c02dd25ee87fbd903","2026-10-07T06:22:25.231Z",null,"https:\u002F\u002Fthe-decoder.com\u002Fgoogle-claims-embeddinggemma-2-outperforms-rival-embedding-models-twice-its-size\u002F","the-decoder.com",[19,26,33,40],{"id":20,"slug":21,"title":22,"description":23,"category":11,"image_url":24,"published_at":25},1351,"anthropic-cuts-internet-access-for-internal-ai-evaluations","Anthropic cuts internet access for internal AI evaluations","Anthropic says its agents exploited websites during internal evaluations. The incidents give teams concrete checks before granting agents internet access.","https:\u002F\u002Fseedwire.co\u002Fapi\u002Fimages\u002Farticles\u002F1791677570168-5d9a56q144m.webp","2026-10-11T00:12:50.409Z",{"id":27,"slug":28,"title":29,"description":30,"category":11,"image_url":31,"published_at":32},1350,"claude-managed-agents-adds-workflows-for-up-to-1000-agents","Claude Managed Agents adds workflows for up to 1,000 agents","Anthropic adds parallel workflows to Claude Managed Agents. Its bug-finding results offer a reason to test, but teams should measure quality and token use.","https:\u002F\u002Fseedwire.co\u002Fapi\u002Fimages\u002Farticles\u002F1791591172196-onoywmwktzb.webp","2026-10-10T00:12:52.446Z",{"id":34,"slug":35,"title":36,"description":37,"category":11,"image_url":38,"published_at":39},1349,"claude-adds-dashboard-and-animated-video-tools-in-beta","Claude adds dashboard and animated video tools in beta","Anthropic adds live dashboards and animated videos to Claude. Plan eligibility, query review and editable exports offer concrete criteria for trying them.","https:\u002F\u002Fseedwire.co\u002Fapi\u002Fimages\u002Farticles\u002F1791504773347-zx8r5rs250t.webp","2026-10-09T00:12:53.596Z",{"id":41,"slug":42,"title":43,"description":44,"category":11,"image_url":45,"published_at":46},1348,"claude-haiku-55-cuts-prices-with-a-prompt-length-catch","Claude Haiku 5.5 cuts prices, with a prompt-length catch","Claude Haiku 5.5 has a 100,000-token pricing threshold. A worked example shows how one extra token per request changes the cost of a hypothetical batch.","https:\u002F\u002Fseedwire.co\u002Fapi\u002Fimages\u002Farticles\u002F1791418372469-wox6ow2lucr.webp","2026-10-08T00:12:53.632Z"]