<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>devtake.dev — Google DeepMind</title><description>Articles on devtake.dev covering Google DeepMind.</description><link>https://devtake.dev/</link><language>en-us</language><item><title>DeepSeek V4 lands: 1.6T-param open MoE, 1M-token context, and SWE-bench within 0.2 of Opus 4.6</title><link>https://devtake.dev/article/deepseek-v4-release/</link><guid isPermaLink="true">https://devtake.dev/article/deepseek-v4-release/</guid><description>DeepSeek shipped V4-Pro and V4-Flash under MIT on April 24. V4-Pro hits 80.6% on SWE-bench Verified. V4-Flash is $0.14 in / $0.28 out.</description><pubDate>Fri, 24 Apr 2026 21:30:00 GMT</pubDate><category>ai</category><category>deepseek</category><category>deepseek-v4</category><category>llm</category><category>ai-models</category><category>open-weights</category><category>moe</category><category>benchmarks</category><category>open-source</category><author>editorial-team</author></item><item><title>Gemini Robotics-ER 1.6 is the brain, not the hands. Here&apos;s how the new two-model robotics stack works.</title><link>https://devtake.dev/article/gemini-robotics-er-1-6/</link><guid isPermaLink="true">https://devtake.dev/article/gemini-robotics-er-1-6/</guid><description>Google DeepMind&apos;s Gemini Robotics-ER 1.6 jumps instrument reading from 23% to 93% and slots in above a separate VLA model. Boston Dynamics&apos; Spot is already running it.</description><pubDate>Sun, 19 Apr 2026 14:00:00 GMT</pubDate><category>ai</category><category>google-deepmind</category><category>gemini-robotics</category><category>embodied-reasoning</category><category>robotics</category><category>boston-dynamics</category><category>vla</category><category>spot</category><author>editorial-team</author></item></channel></rss>