由Google云公司赞助
选择您的第一个基因AI 使用案例
要从基因AI开始，首先关注能够改善人类信息体验的领域。
削减意味着企业对成本较低的模型有更多的选择，但许多人仍然难以决定用于特定用途的模型。
3 小读
Justin Sullivan通过 Getty 图像
OpenAI的最新型号GPT-6.1 Terra比之前的版本便宜了50%，显示了基因AI供应商如何继续应对开放重量供应商的价格压力。 然而，升级的AI模型价格战对目前以最低价格面对多种模型的企业来说可能变得令人生畏。
OpenAI 推出周一更新型号，发布GPT-6.1 Sol后近一个月。 新的Terra模型快速高效，结合了视觉推理和精确使用计算机，OpenAI 称。 该模型擅长计算机任务，如起草建议书，填表和编辑视频。
然而，其价格是其真正的销售点：每百万美元输入令牌1美元，每百万产出令牌5美元。 相比之下，GPT-6.1 索尔每百万个输入令牌花费2美元，每百万个输出令牌花费10美元。 GPT-5.6 Terra也花费每百万个输入符2美元，每百万个输出符12美元。
GPT-6.1 Terra的发布标志着OpenAI连续第四次削价其型号，这一趋势及其主要竞争对手Anthropic在与Alibaba,Baidu和Z.ai等中国小贩的开放量产型号竞争时正在扩展。 相比之下，Anthropic本周向证监会提交了IPO peopleus，也大幅降低了定价，最近的一次降价是在10月，当时供应商将Haiku 5.5的价格降低了90%/100K令牌和每百万产出令牌0.5美元。
称 David Nicholson是Futurum集团的分析师，“如果一切都以最高价格计价，那么人们就会寻找其他选择，因此，他们试图做的就是尽可能提高工具的效率。 ”
同时，对于模型用户来说，即使模型制造者继续降低价格，也不清楚哪个模型是正确的环境或工作量。
“我们距离企业能够有效地利用最符合成本效益的模式完成它们所希望完成的每一项任务还有很长的路要走。” “有更大的选择是件好事
尽管更便宜的模型让企业能够优化象征性消费和他们正在做的工作的成本，但“与之相关的是摩擦，因为你必须具备一定程度的理解和洞察力，了解模型适合什么，以及你要求模型如何与该模型适合什么，而且并不总是100%清晰。
此外，对于OpenAI和Anthropic来说，竞相降低模型价格可能会产生明显的后果，特别是因为这两家公司都追求公开上市，尼科尔森称。 降低价格可能导致投资者认为技术价值较低，或表明需求增加，因此利润增加。
“这还有待观察”,Nicholson 称。
除了降低价格之外，OpenAI 称 GPT-6.1 Terra在一些任务上还拥有相对于其Astra和Sol模型的耐久优势。 这意味着它可以加速更多的代理，高容量的企业工作流程，称钱德拉塞卡兰是加特纳的一位分析师。 然而，加速企业工作流程需要更加谨慎，he 称。
Chandrasekaran补充说，“更大的自主权也增加了业务和安全风险，使模式路线、任务层面的ROI、准入控制和人员监督对安全推广使用至关重要。”
Esther Shittu是一名TechTalget新闻编剧和播客制作人和主持人，报道AI技术。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
The cut means enterprises have more options for lower-cost models, but many still struggle to decide which model to use for particular use cases.
3 Min Read
Justin Sullivan via Getty Images
OpenAI’s latest model, GPT-6.1 Terra, is 50% cheaper than its previous version, showing how the generative AI vendor continues to respond to price pressure from open-weight providers. However, the escalating AI model price war could become daunting for enterprises now faced with multiple models at their lowest prices.
OpenAI introduced the updated model on Monday, nearly a month after releasing GPT-6.1 Sol. The new Terra model is fast and efficient, combining visual reasoning and precise use of computers, OpenAI said. The model is good at computer-based tasks such as drafting proposals, filling out forms and editing videos.
However, its price is its true selling point: $1 per million input tokens and $5 per million output tokens. In comparison, GPT-6.1 Sol costs $2 per million input tokens and $10 per million output tokens. GPT-5.6 Terra also costs $2 per million input tokens and $12 per million output tokens.
The release of GPT-6.1 Terra marks the fourth consecutive time OpenAI has slashed the price of its models, a trend it and its chief rival Anthropic are extending as they compete against open-weight models from Chinese vendors such as Alibaba, Baidu and Z.ai. Comparatively, Anthropic, which filed its IPO prospectus with the SEC this week, has also slashed its pricing, with the most recent price cut in October, when the vendor reduced the price of Haiku 5.5 by 90% per 100K tokens and $0.50 per million output tokens.
“They are refining our ability to get the most bang for our buck out of these models,” said David Nicholson, an analyst at Futurum Group. “If everything is priced at the highest level, then people will look for alternatives, so what they’re trying to do is make their tools as efficient as possible.”
Meanwhile, for model users, it’s unclear which model is right for which environment or workload, even as model makers continue to reduce prices.
“We are a long way from enterprises being able to efficiently leverage the most cost-effective model for each task that they want achieved,” Nicholson said. “It’s great to have more choice, but I think the choice is causing confusion.”
While cheaper models enable enterprises to optimize token consumption and cost for the work they’re doing, “there’s a friction associated with it because you have to have a certain level of understanding and insight into what the models are suited for and how what you’re asking the model to do lines up with what that model is suited for, and it is not always 100% clear, he added.
Moreover, for both OpenAI and Anthropic, the race to lower model prices could have distinct consequences, especially as both companies pursue public listings, Nicholson said. Price reductions could lead investors to think the technology is less valuable, or it could indicate greater demand and, therefore, greater profitability.
“It remains to be seen,” Nicholson said.
Other than reducing the price, OpenAI said GPT-6.1 Terra also has a latency advantage over its Astra and Sol models on some tasks. This means it can accelerate more agentic, high-volume enterprise workflows, said Arun Chandrasekaran, an analyst at Gartner. However, speeding up enterprise workflows requires greater caution, he said.
“Greater autonomy also increases operational and security risks, making model routing, task-level ROI, access controls, and human oversight essential to scaling adoption safely,” Chandrasekaran added.
Esther Shittu is a TechTarget news writer and podcast producer and host covering AI technology.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.