Skip to content
TheChatArchiveThe Chat Archive
Geminigemini-2.5-pro-exp-03-25-grounding· GoogleHugging Face LMArena Search Arena 24K

patent for LEGO bricks expired in which year?

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with gpt-4o-search-preview’s answer.
  1. User

    patent for LEGO bricks expired in which year?

  2. Gemini Assistant

    The patent for the LEGO brick has expired, but sources offer different years for when this occurred:

    • 1978: One source indicates that the last significant patent for the standard LEGO brick expired in 1978.[[1]] A specific US patent (US3005282A), filed in 1958 and granted in 1961, shows an anticipated expiration date of October 24, 1978.[[2]]
    • 1988: Another source mentions the expiry of the LEGO brick patent occurred in 1988.
    • 2011: Some sources state that the original patents expired in 2011. This might refer to later patents or patents in different jurisdictions.

    Key Points:

    • Patents typically last for 20 years and cannot be extended.[[3]]
    • The original patent application for the modern LEGO brick design was filed in Denmark on January 28, 1958.[[1]][[2]] A corresponding US patent was filed later that year.[[2]]
    • After the patent expired, LEGO attempted to protect the brick's design through trademark law, registering it as a 3D trademark.[[4]] However, these efforts faced legal challenges and were largely unsuccessful because the brick's features (like the studs) were deemed functional rather than purely distinctive design elements.
    • While the basic brick patent expired, LEGO continues to hold thousands of patents for other inventions, including electrical components, specialized pieces, minifigure features, and gaming systems. They also rely on copyright and design rights for aspects like construction kit designs and minifigures.[[3]]
  3. User

    那为啥近年来还是有侵权案件?

  4. Gemini Assistant

    您问得非常好!虽然乐高(LEGO)最核心的基础积木块的专利确实早已过期(不同来源指出是1978年、1988年或2011年),但这并不意味着其他公司可以随意复制乐高的一切产品而不承担法律责任。

    近年来仍然有侵权案件发生,主要是因为乐高除了基础砖块专利外,还拥有大量的其他知识产权(IP),这些知识产权仍在保护期内,具体包括:

    1. 商标权 (Trademark):

      • 品牌名称和Logo: “LEGO”这个名字和其标志性的红色背景白色字母Logo是注册商标。未经授权使用这些商标进行销售(例如,在假冒产品上使用LEGO标志)是商标侵权。
      • 服务商标: 有些案例涉及假冒乐高授权店,在店铺招牌、装修、宣传等方面使用乐高商标提供教育培训服务,这也构成商标侵权。
      • 立体商标 (3D Trademark): 乐高曾尝试将积木块的形状注册为立体商标,但因其功能性太强而在很多地方失败。不过,对于乐高小人仔(Minifigure)的立体形状,乐高仍在积极寻求并维护其商标权,尽管这也面临争议。
    2. 著作权 (Copyright):

      • 美术作品: 乐高套装的包装盒设计、说明书的图纸和内容、积木拼搭成的具体立体造型、以及乐高小人仔的设计,都被认为是美术作品,享有著作权。
      • 衍生作品: 乐高基于其产品推出的主题动画片、书籍等周边产品中的角色和故事也受著作权保护。
      • 侵权行为: 很多侵权案件,如著名的“乐拼”案和“博乐”案,都涉及大量、几乎1:1地复制乐高玩具的包装、说明书、立体造型和小人仔设计。这些行为被认定为侵犯了乐高的著作权。因为复制的是受版权保护的“表达形式”,即使使用了自己的商标(如“乐拼”、“博乐”),仍然构成侵权,情节严重时甚至构成刑事犯罪。
    3. 新专利和设计权 (New Patents and Design Rights):

      • 虽然基础砖块专利过期,但乐高不断研发新的、特殊的积木零件、连接件、功能组件(如气动、电动零件)、小人仔配件等,并为这些新设计申请新的专利或外观设计保护。复制这些仍在保护期内的新设计,会构成专利或设计侵权。
    4. 反不正当竞争法 (Unfair Competition Law):

      • 即使不直接侵犯上述明确的知识产权,如果一家公司的行为(如生产销售与乐高产品高度相似、足以引起消费者混淆的产品,特别是模仿乐高知名系列产品的特有包装装潢)违反了诚实信用原则,不正当地利用了乐高的市场声誉,也可能构成不正当竞争。一些判决中,法院认定乐高的某些系列玩具(如“幻影忍者”)的包装装潢属于知名商品特有的包装装潢,受到反不正当竞争法的保护。

    总结来说:

    基础砖块专利过期意味着其他厂商可以生产与乐高积木兼容的、尺寸相同的砖块。但是,这并不等于可以随意复制乐高的:

    • 品牌名称和Logo
    • 具体的套装设计、拼搭说明书、包装艺术
    • 受版权保护的小人仔形象
    • 仍在专利或设计权保护期内的新型特殊零件
    • 会引起市场混淆的整体产品外观和包装

    因此,近年来的侵权案件大多是围绕着著作权、商标权以及不正当竞争行为展开的。乐高集团在中国及全球范围内一直积极通过法律途径打击

  5. User

    The best LLM on lmarena today?

  6. Gemini Assistant
    1. Synthesize the search results:

      • The LMSys Chatbot Arena ranks LLMs based on crowdsourced, blind, head-to-head comparisons using an Elo rating system.
      • The leaderboard is frequently updated.[[1]] Search results indicate updates as recent as March 25th or 26th, 2025 ([2], [4]).
      • Results from 2023 ([1], [11]) and mid-2024 ([6]) are outdated regarding the top models but explain the methodology.
      • Results from late 2024 ([8]) and early 2025 ([5], [10]) mention models like GPT-4-Turbo, Claude 3 Opus, Gemini 1.5 Pro, Qwen2.5-Max, and Llama 3 70b as top contenders at those times.
      • Several sources ([2], [3], [4], [7], [10]) point to lmarena.ai or a Hugging Face space as the current home for the live leaderboard.
      • The key is to find the most recent ranking. The search results show updates around March 25th/26th, 2025 ([2], [4]). While the results don't explicitly list the number 1 model today (March 30th, 2025), they confirm the dynamic nature and point to where the live ranking can be found.
      • One result ([8] from Nov 2024) listed GPT-4-Turbo as #1 at that time. Another ([5] from Feb 2025) mentioned Qwen2.5-Max was #7 overall but #1 in math/coding.[[2]] Another ([6] from June 2024) highlighted GPT-4o, Gemini 1.5, and Claude 3 Opus as very close top performers then.[[3]]
      • The most recent update mentioned is March 25th/26th, 2025 ([2], [4]). Since specific rankings for today aren't in the static results, the best approach is to state this and point the user to the live leaderboard.
    2. Construct the answer:

      • Start by explaining what the LMSys Chatbot Arena is and how it ranks models (Elo rating based on human votes).
      • Emphasize that the rankings are dynamic and change frequently. Mention the most recent update date found in the search results (around March 25/26, 2025).
      • Since the search results don't provide the exact live ranking for today (March 30, 2025), state that directly.
      • Mention the models that have recently held top positions or were highly ranked according to the search results (e.g., GPT-4 models, Claude 3 Opus, Gemini 1.5 Pro, potentially newer models like Qwen2.5-Max in specific categories).
      • Provide the user with the direct link to the official Chatbot Arena leaderboard (lmarena.ai or the Hugging Face space) so they can see the absolute latest ranking.
      • Optionally, differentiate between overall rankings and category-specific rankings (like coding, math, hard prompts) as mentioned in sources [4], [5], [9].
    3. Refine the answer: Ensure clarity, accuracy based only on the provided search snippets, and directness. Add a disclaimer about the dynamic nature of the leaderboard. Make sure to use the current date (March 30, 2025).Okay, to find the best LLM on the LMSys Chatbot Arena today (Sunday, March 30, 2025), we need to look at their frequently updated leaderboard.

    The Chatbot Arena uses an Elo rating system, similar to chess rankings, based on thousands of blind, head-to-head votes from human users comparing anonymous model outputs ([2], [3], [4], [10], [11]). This leaderboard is dynamic and reflects ongoing user preferences.

    Based on the search results:

    1. The leaderboard is updated frequently: The most recent updates mentioned in the search results are around March 25th or 26th, 2025 ([2], [4]).
    2. Top Contenders (Historically & Recently): While the absolute current number one isn't explicitly stated in the static search results for today, models that have recently held top spots or been highly competitive include ([5], [6], [8]):
      • GPT-4 models (like GPT-4-Turbo, GPT-4o, various preview versions) from OpenAI.[[4]]
      • Claude 3 models (especially Opus) from Anthropic.[[3]][[4]]
      • Gemini models (like Gemini 1.5 Pro) from Google.[[4]]
      • Llama 3 models (like 70b Instruct) from Meta (often the top open-source model) ([8]).[[4]]
      • Qwen models (like Qwen2.5-Max) from Alibaba Cloud, noted for strong performance, particularly in technical areas like math and coding ([5], [10]).[[2]]

    Where to find the current ranking:

    Since the leaderboard changes, the best way to know the exact top model right now is to check the official sources directly:

    • The official Chatbot Arena website: lmarena.ai ([2], [3], [7])
    • The Hugging Face Space mirror: (Search results [2] and [3] mention a Hugging Face space hosting the leaderboard)

    In summary: As of the last updates found (late March 2025), models from OpenAI (GPT-4 series), Anthropic (Claude 3 Opus), and Google (Gemini 1.5 Pro) are typically vying for the top spots in the overall rankings. However, you should check the live leaderboard at lmarena.ai for the definitive ranking today.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.