快来看,n8n更新了!试用新模型和服务,无需设置账户

内容来源:https://blog.n8n.io/gateway-credits/
内容总结:
n8n推出Gateway积分功能,用户无需注册第三方账户即可直接调用AI模型和服务
n8n近日宣布推出Gateway积分(Gateway credits)功能,旨在解决用户在自动化工作流中调用第三方AI服务时面临的账户注册和API密钥配置难题。此前,用户在n8n中使用不熟悉的节点时,往往需要先前往服务提供商页面注册账户、绑定支付方式、生成API密钥并配置凭证,这一流程在时间紧迫或仅作探索时足以让一个有潜力的想法搁浅。
现在,在n8n Cloud自助服务计划中,用户添加受支持的AI模型或服务节点时,只需选择“Gateway积分”而非自行添加提供商凭证,即可直接运行工作流。使用量从n8n中的共享预付费余额中扣除,无需创建额外账户或管理更多API密钥。对于允许团队共享余额的组织而言,这意味着更少的供应商账户、发票和报销流程。
在定价方面,n8n表示其目标是遵循各提供商公布的标价,并将所有费率集中在一个公开页面上供用户随时查阅。余额以真实货币显示,Cloud管理面板支持按工作流或按服务查看使用情况。n8n强调,此举并非为便利性收取溢价,而是为用户提供一条通往实用工具的最短路径。如果用户后续希望与提供商建立直接关系,仍可像以往一样使用自己的API密钥。
此外,Gateway积分同样适用于通过n8n Assistant使用自然语言提示生成的工作流。无论是从模板、空白画布还是提示词开始,用户从想法到首次成功运行的步骤都大幅减少。
目前,Gateway积分支持六家模型提供商和五项工具服务:
模型提供商包括Anthropic、Google Gemini、OpenAI等知名提供商,以及以下开放模型:
- 阿里云(Qwen Cloud节点):利用全球最大的开源模型家族构建面向复杂企业、研究和长周期任务的编程智能体。
- Kimi.ai:将长周期编程和研究任务交给专为多步骤任务设计的开放模型,支持思考模式和即时模式。
- MiniMax:使用M3进行100万token上下文内的编程和智能体工作,或使用H3进行全模态视频生成。
工具服务包括:
- Brave Search:基于其自有400亿页面索引为AI应用提供实时搜索结果,无需Google抓取,无上游依赖,结构化的零数据保留,定价简单实惠。
- Browserbase:为工作流提供浏览器智能体,可导航、点击、输入并从网站提取结构化数据,然后将结果传入自动化流程或智能体的其余部分。
- Firecrawl:将网站转化为LLM可用数据,支持抓取动态页面、解析媒体和爬取整个站点,自动处理JavaScript、速率限制和代理。
- LlamaParse:智能体OCR,将复杂企业文档转化为AI可用的上下文和结构化数据,在成本与准确性方面处于帕累托前沿。
- PDF.co:在单个节点中生成、转换和提取PDF,支持HTML转PDF、PDF转结构化数据、合并、拆分和填写。
n8n表示,这十一个提供商和服务仅是起点,未来名单将持续扩展。
在实际应用场景中,n8n以一个竞争情报工作流为例展示了Gateway积分的价值:用户可直接在画布上构建或通过自然语言向n8n Assistant描述工作流。以Brave Search查找近期产品发布和新闻报道为起点,Browserbase处理需要交互才能加载内容的页面,LlamaParse解析复杂的产品简报和分析报告PDF,最后选择任一受支持模型汇总发现、对比上周报告并起草更新。整个过程无需离开n8n编辑器即可完成。
对于知识库和研究智能体,Firecrawl可爬取整个站点并返回LLM可用的markdown或结构化数据;对于文档工作流,PDF.co可在单个节点中生成精美PDF、提取结构化数据或合并、拆分和填写文件。
在管理方面,Gateway积分来自n8n实例中共享的单一余额,用户只需关注一个数字、在一个地方充值。通过Cloud管理面板,可按服务和工作流跟踪使用情况、添加资金或设置自动充值。共享余额还使团队能够更自由地选择最适合工作流各环节的模型或工具,而不必因为凭证已配置就默认使用某一服务。
n8n透露,目前正在开发下一批精选服务,重点关注用户最常使用的API和能力。此外,还在探索如何让Gateway积分支持更多提供商,并让团队对实例内积分的管
中文翻译:
在n8n中尝试一个模板或搭建一个工作流,有时会意外地被引到某个服务提供商的注册页面。在你还没来得及了解一个陌生节点能为自动化带来什么之前,可能就需要先创建账户、添加支付方式、生成API密钥,并在n8n中配置好凭证。当你时间紧张或只是想随便试试时,这一停顿就足以让一个原本很有前景的想法还没跑起来就被搁置了。
现在,创建新账户不再需要出现在看到绿色对勾之前。当你在n8n Cloud自助服务方案中添加受支持的AI模型或服务节点时,运行所需的一切都已经在n8n中准备就绪。只需选择Gateway积分,而无需添加你自己的提供商凭证。用量从n8n中的共享预付费余额中扣除,因此你可以直接开始工作,无需再创建另一个账户或管理另一个API密钥。对于允许这样做的团队来说,一个共享余额还意味着需要管理的供应商账户、发票和费用报告都更少。
我们还希望成本方面让人省心——是好的那种省心。我们的目标是:凡是提供商有公开标价的,就遵循其标价,并把所有费率放在一个公开页面上,让你在运行任何东西之前都能查看。你的余额以真实货币显示,Cloud管理面板让你可以按工作流或按服务查看使用情况。我们这样做不是为了因便利而加价。这是我们能找到的、让你和今天就能派上用场的工具之间最短的路径。如果你之后决定想与提供商建立直接关系,使用自己的密钥的方式和以前完全一样。
你还可以在使用n8n Assistant时,为受支持的节点使用Gateway积分,从自然语言提示直接生成可用的自动化。无论你是从模板、空白画布还是提示开始,你的想法到第一次成功运行之间的步骤都更少了。
今天可以在哪里使用Gateway积分
能够首次运行只是其中一部分。Gateway积分还让你更容易对已在使用的 workflow 进行实验,无论是尝试不同的模型,还是添加一个你一直好奇的工具。
目前有六个模型提供商和五个工具服务支持Gateway积分。
模型提供商
可以使用一些最知名的提供商,如Anthropic、Google Gemini和OpenAI,也可以试试以下这些你可能还没机会使用的开放模型。
- 阿里云(Qwen Cloud节点):利用全球最大的开源模型家族,为复杂的企业级、研究和长周期任务构建编程智能体。
- Kimi.ai:将长周期编程和研究任务交给一个开放模型,它专为在多步骤任务中保持正轨而构建,并根据任务类型提供思考模式和即时模式。
- MiniMax:使用M3在100万token上下文内进行编程和智能体工作,或使用H3进行全模态视频生成。
工具服务
- Brave Search:让AI应用基于其自有400亿页面索引的实时结果——无需抓取Google,无上游依赖,结构化零数据保留,定价简单且实惠。
- Browserbase:为你的工作流提供浏览器智能体,能够导航、点击、输入,并从网站提取结构化数据,然后将结果传入自动化或智能体的其余部分。
- Firecrawl:通过抓取动态页面、解析媒体和爬取整个网站,将网站转化为LLM就绪的数据。Firecrawl为任何AI工作流处理JavaScript、速率限制和代理。
- LlamaParse:智能体OCR,将复杂的企业文档转化为AI就绪的上下文和结构化数据,在成本和准确率上达到帕累托前沿。
- PDF.co:在单个节点中生成、转换和提取PDF。HTML转PDF、PDF转结构化数据、合并、拆分和填写。
这十一个提供商和服务只是我们的起点,而非终点。预计这个列表会持续增长。
这在实践中能带来什么
拿一件普通的事来说:一个每周交付有用更新的竞争情报工作流。
你可以直接在画布上构建,或用自然语言向n8n Assistant描述这个工作流。使用n8n Assistant时,如果你指定了一个受支持的服务且尚未拥有其凭证,它会自动应用Gateway积分。
对于这个工作流,可以从Brave Search开始,它能在实时网络上找到最近的产品发布、新闻报道和其他变化。结果来自Brave自有的独立索引,为工作流提供最新来源,可直接传入后续的研究和分析环节。
搜索能帮你完成大部分工作,但有些有用的细节可能藏在交互式定价页面或更新日志中,只有页面加载后或有人点击后才会显示内容。Browserbase可以在托管的云浏览器中打开这些页面,进行导航、点击、输入,并提取工作流所需的信息。这项复杂且耗时的任务——深度研究往往需要它——只需添加节点就能快速完成。
然后是搜索找到的产品简报、分析师报告和PDF。一个基础的提取器可能会把精心结构的表格压扁成面目全非的东西,让下游模型只能靠猜。LlamaParse结合了布局感知的计算机视觉、OCR和视觉语言模型,将这些复杂文档解析为结构化内容,保留更多表格、标题和布局——正是这些东西赋予了信息意义。
最后,选择任何受支持的模型来汇总发现,与上周的报告进行比较,找出变化,并起草更新。加上你的独特参数,你就获得了来自真正专业提供商的全面竞争情报,全程无需离开n8n编辑器去创建新账户。
这只是组合首发可用服务的一种方式。对于知识库和研究智能体,Firecrawl可以爬取整个网站并返回LLM就绪的markdown或结构化数据。对于文档工作流,PDF.co可以在一个节点中将生成的内容转化为精美的PDF、提取结构化数据,或合并、拆分和填写文件。
更多选择,统一余额管理
一旦你运行的工作流超过几个,甚至只是一个复杂的工作流,提供商分散就可能变成一项自己的维护工作。设置账户、跟踪多个仪表盘、管理费用报告,都会增加工作量。Gateway积分来自你n8n实例共享的一个余额,让你只需关注一个数字、在一个地方充值。在Cloud管理面板中,你可以按服务和工作流跟踪用量、充值,或设置自动充值。
跨服务的共享余额也让你和你的团队更自由地选择最适合工作流每个部分的模型或工具,而不是因为凭证已经配好了就默认使用某一个。试试你一直好奇的模型,或为你已经在运行的工作流添加一个新服务。你将能够在承诺开设另一个提供商账户之前,先了解它对工作流的影响。
下一步走向
六个模型提供商和五个工具服务只是起点。我们已经在筹备下一批精选服务,聚焦人们最常用的API和能力。除此之外,我们正在探索Gateway积分如何支持更多提供商,并让团队对实例内积分的管理方式拥有更多控制权。
Gateway积分现已在n8n Cloud Starter和Pro方案中可用。我们也希望随着时间推移将同样的体验带到自托管实例,但这需要额外的基础设施,不属于本次发布的范围。
你的n8n Cloud账户可能已经有Gateway积分了
符合条件的新Cloud账户(包括试用账户)初始会获得一小笔Gateway积分余额。现有的付费Starter和Pro Cloud客户在推广期间也收到了一次性额度。
把它花在你一直想尝试的东西上吧。用OpenAI、Anthropic或Google Gemini的前沿模型来实践一个新想法。如果你一直对开放模型好奇,看看Kimi.ai、阿里云或MiniMax能在你已经在运行的工作流中做些什么。如果你想让智能体更好地访问网络,试试Brave Search、Firecrawl或Browserbase。如果你的工作流卡在文档处理上,试试LlamaParse或PDF.co。
无论你选择什么,Gateway积分都让你无需离开画布就能更轻松地让新的模型或服务投入使用。
英文来源:
Trying a template or building a workflow in n8n can sometimes lead you on an unexpected detour to a service provider’s signup page. Before you can find out what an unfamiliar node adds to your automation, you may need to create an account, add a payment method, generate an API key, and set up the credential in n8n. When you’re short on time or simply exploring, that pause can be enough to keep a promising idea from getting its first run.
That new account creation no longer has to come before the satisfying sight of green checkmarks. Now, when you add a node for a supported AI model or service on self-service n8n Cloud plans, everything you need to run it is already available in n8n. Just select Gateway credits instead of adding your own provider credentials. Usage comes from a shared prepaid balance in n8n, so you can get to work without creating another account or managing another API key. For teams that allow it, one shared balance can also mean fewer vendor accounts, invoices, and expense reports to manage.
We also wanted the cost side to be boring in a good way. Our aim is to follow each provider’s published list price wherever one exists and put every rate on a single public page you can check before you run anything. Your balance is shown in real currency, and the Cloud Admin Panel lets you see how it is being used by workflow or by service. We didn’t build this to add a premium for the convenience. It's the shortest path we could find between you and a tool that does something useful today. If you later decide you’d like a direct relationship with the provider, bringing your own key works exactly like it always has.
You can also use Gateway credits for supported nodes using n8n Assistant to move from a natural-language prompt to a working automation. Whether you start with a template, a blank canvas, or a prompt, there are now fewer steps between your idea and its first successful run.
Where you can use Gateway credits today
Getting to a first run is only part of it. Gateway credits also make it easier to experiment with workflows you already use, whether that means trying a different model or adding a tool you’ve been curious about.
Six model providers and five tool services are available with Gateway credits today.
Model providers
Reach for some of the most recognized providers like Anthropic, Google Gemini, and OpenAI, or try one of these open models you might not have had a chance to use yet.
- Alibaba Cloud (Qwen Cloud Node): Build coding agents for complex enterprise, research, and long-horizon tasks with the world's largest open-source model family.
- Kimi.ai: Hand long-horizon coding and research tasks to an open model built to stay on track across many steps, with thinking and instant modes depending on the job.
- MiniMax: Use M3 for coding and agentic work across a 1M-token context, or H3 for omni-modal video generation.
Tool services - Brave Search:Grounds AI apps in real-time results from its own 40-billion-page index — no Google scraping, no upstream dependencies, structural ZDR, and simple, affordable pricing.
- Browserbase: Give your workflows browser agents that navigate, click, type, and extract structured data from websites, then pass the results into the rest of an automation or agent.
- Firecrawl: Turn websites into LLM-ready data by scraping dynamic pages, parsing media, and crawling entire sites. Firecrawl handles JavaScript, rate limits, and proxies for any AI workflow
- LlamaParse: Agentic OCR that turns complex enterprise documents into AI-ready context and structured data at the Pareto frontier of cost and accuracy.
- PDF.co: Generate, convert, and extract PDFs in a single node. HTML to PDF, PDF to structured data, merge, split, and fill.
These eleven providers and services are where we’re starting, not where we’re stopping. Expect the list to keep growing.
What this gets you in practice
Take something ordinary: a competitive intelligence workflow that delivers a useful update each week.
Start by building on the canvas directly or describing the workflow to n8n Assistant in natural language. When using n8n Assistant, it will apply Gateway credits if you name a supported service and don’t already have a credential for it.
For this workflow, that might begin with Brave Search, which finds recent product announcements, news coverage, and other changes across the live web. Results come from Brave’s own independent index, giving the workflow current sources it can pass directly into the rest of the research and analysis.
Search gets you most of the way, but some useful details may sit on an interactive pricing page or a changelog that only reveals its content after the page loads or someone clicks through it. Browserbase can open those pages in a managed cloud browser, navigate, click, type, and extract the information your workflow needs. This complex and time-consuming task that is often required for in-depth research, is handled quickly by adding the node.
Then there are the product briefs, analyst reports, and PDFs that the search turns up. A basic extractor can flatten a carefully structured table into something unrecognizable, leaving the model downstream with guesswork. LlamaParse combines layout-aware computer-vision, OCR, and vision-language models to parse these complex documents into structured content that preserves more of the tables, headings, and layouts that give the information meaning.
Finally, choose any of the supported models to bring the findings together, compare them with last week’s report, identify what changed, and draft the update. Add your unique parameters and you’ve got comprehensive competitive intelligence from real expert providers without ever leaving your n8n editor to set up a new account.
That is only one way to combine the services available at launch. For knowledge bases and research agents, Firecrawl can crawl an entire site and return LLM-ready markdown or structured data. For document workflows, PDF.co can turn generated content into polished PDFs, extract structured data, or merge, split, and fill files in one node.
More choice, managed in a single balance
Once you’re running more than a handful of workflows, or even one complex one, provider sprawl can become its own maintenance job. Setting up accounts, tracking multiple dashboards, and managing expense reports all add work. Gateway credits come from one balance shared across your n8n instance, giving you one number to watch and one place to top up. From the Cloud Admin Panel, you can track usage by service and workflow, add funds, or set up automatic top-ups.
A shared balance across services also gives you and your team more freedom to choose the model or tool that fits each part of a workflow best, instead of defaulting to one because its credentials are already in place. Try a model you’ve been curious about, or add a new service to a workflow you already run. You’ll be able to find out how it affects your workflow before committing to another provider account.
Where this goes next
Six model providers and five tool services are only the starting point. We’re already working on the next curated batch, focusing on the APIs and capabilities people use most. Beyond that, we’re exploring how Gateway credits can support many more providers and give teams more control over how credits are managed across an instance.
Gateway credits are available on n8n Cloud Starter and Pro today. We also want to bring the same experience to self-hosted instances over time, though that requires additional infrastructure and is not part of this release.
Your n8n Cloud account may already have Gateway credits
Eligible new Cloud accounts, including trials, start with a small balance of Gateway credits. Existing paying Starter and Pro Cloud customers also received a one-time allowance during the rollout.
Spend it on something you’ve wanted to try. Put a frontier model from OpenAI, Anthropic, or Google Gemini to work on a new idea. If you’ve been curious about open models, see what Kimi.ai, Alibaba Cloud, or MiniMax can do in a workflow you already run. If you want to give an agent better access to the web, try Brave Search, Firecrawl, or Browserbase. If documents are where your workflow gets stuck, try LlamaParse or PDF.co.
Whatever you choose, Gateway credits make it easier to put a new model or service to work without leaving the canvas.
文章标题:快来看,n8n更新了!试用新模型和服务,无需设置账户
文章链接:https://news.qimuai.cn/?post=5096
本站文章均为原创,未经授权请勿用于任何商业用途