谷歌 Deepmind 推出 Gemini Robotics ER 2 模型
Related
Google DeepMind launches robotics accelerator program in Europe, with 16 startups selected for the first batch.
On June 9th, Google DeepMind announced the launch of its robotics accelerator program, targeting early-stage robotics startups in Europe. The three-month program saw its first cohort of participants gather in London this week. Selected companies will receive mentorship from Google DeepMind technical experts, access to Gemini robot models, and support from Google's AI technology stack. The initial 16 selected companies cover multiple fields including logistics, manufacturing, healthcare, construction, marine exploration, and neurosurgery. (Jiemian)
GPT-5.6, Gemini 3.5 Pro, and Grok 4.5 will all be released soon.
PANews reported on July 7th that, according to market sources, Gemini 3.5 Pro will be officially released on July 17th. Its front-end and visual code generation capabilities are said to have seen a significant leap forward, outperforming Anthropic's Fable 5 in multiple tests. However, it still lags behind its competitor in hardcore inference and complex engineering tasks. Furthermore, Google DeepMind has abandoned the original 2.5 Pro platform, opting instead for completely new pre-training on Gemini 3.5 Pro, thus delaying the release date from the original June 2026 to July 17th. Elon Musk previously announced that xAI's latest large model, Grok 4.5, has begun beta testing within SpaceX and Tesla.
GPT-5.6 is rumored to be available next week, while the Gemini 3.5 Pro, featuring 2M context, will be released later.
According to Beating, tech blogger leo revealed that OpenAI may release GPT-5.6 to the public between July 7th and 9th, with the earliest possible release date being July 7th. The new model package will have more lenient pricing, and OpenAI has also strengthened its security measures before the launch. Google DeepMind's Gemini 3.5 Pro is rumored to be tentatively scheduled for release on July 17th. Another blogger, Astro Polo, claims that Gemini 3.5 Pro will support a 2 million token context window, double the current 1 million token context window of Claude Sonnet 5, Claude Opus 4.8, and Claude Fable 5, making it more suitable for handling long codebases, large documents, and long conversations.
Google's video model reclaims the top spot: Gemini Omni Flash tops Video Arena rankings.
According to Beating, Google DeepMind's newly released video generation model, Gemini Omni Flash, topped the Video Arena blind test leaderboard with an Elo score of 1404. This new model leads ByteDance's Seedance 2.0 Mini, which previously held the top spot, by a full 101 points, setting a new record for the largest score difference on this leaderboard. Video Arena, a leaderboard based on blind human voting, has previously been dominated by ByteDance's Seedance series models in the top three. Seedance 2.0 Mini, with its better multi-camera stability and faster generation speed, ranked first with 1303 points. Gemini Omni Flash's rise to the top signifies that Google's video generation model has overtaken competitors like ByteDance within six months, leaping from a lagging position during the Veo era to the forefront.
Reports suggest that the release of GPT-5.6 and Gemini 3.5 Pro has been delayed until July, while OpenAI's new speech model may be launched this week.
According to a post by tech blogger @synthwavedd on Odaily Odaily, several highly anticipated cutting-edge AI models have had their release schedules adjusted. GPT-5.6, originally scheduled for release soon, has been postponed to mid-July, while Google DeepMind has cancelled the planned late June release of Gemini 3.5 Pro due to dissatisfaction with the model's current performance. Meanwhile, preparations for the launch of OpenAI's next-generation bidirectional speech model, Bidi, are underway on the ChatGPT platform, and it may be available to users as early as this week. Reports indicate that Bidi supports full-duplex voice interaction, allowing users and the model to speak simultaneously and interrupt the conversation at any time, and is considered a significant upgrade to existing voice models. In addition, Anthropic has granted early access to the Claude Sonnet 5 to select enterprise customers. Given the slowdown in the release of its flagship models, the Mythos 5 and Fable 5, the Claude Sonnet 5 is seen as a temporary product solution for Anthropic to alleviate competitive pressure.
Nobel laureate John Jumper leaves DeepMind; Google loses two AI executives in two days.
According to Beating, John Jumper, Vice President of Google DeepMind and head of the AlphaFold team, announced he will be joining Anthropic. Jumper worked at Google DeepMind for nearly nine years and, along with CEO Demis Hassabis, was awarded the 2024 Nobel Prize in Chemistry for leading the development of AlphaFold. Just the day before, Noam Shazeer, former co-head of Gemini and co-author of "Attention Is All You Need," also announced his departure from Google to join OpenAI as head of architecture research.