OpenAI's New GPT-4o Mini: Affordable, Improved AI Model with Advanced Capabilities and Expanded Data Analysis on Azure

Redmond, Washington United States of America
Advanced data analysis capabilities allow for real-time updates and custom GPTs
GPT-4o mini offers expanded context window, multilingual capabilities, and safety features
Model delivers 82% score on MMLU compared to GPT-3.5 Turbo's 70% and is more than 60% cheaper
OpenAI releases GPT-4o mini, an affordable and improved AI model on Azure
OpenAI's New GPT-4o Mini: Affordable, Improved AI Model with Advanced Capabilities and Expanded Data Analysis on Azure

OpenAI, a leading artificial intelligence (AI) research laboratory, has recently released an upgraded version of its model called GPT-4o mini. This new model is more cost-effective and offers improved capabilities compared to the previous version, GPT-3.5 Turbo.

GPT-4o mini is now available on Azure AI, Microsoft's cloud computing service. With this integration, Azure AI provides data residency in all 27 regions and worldwide pay-as-you-go availability for the new model.

According to OpenAI, over 53,000 customers are currently using Azure AI to deliver innovative experiences at scale. Among these customers are more than 50% of Fortune 500 companies that build applications with Azure OpenAI Service.

GPT-4o mini offers text processing capabilities and is expected to support image, audio, and video in the future. It delivers an expanded context window of 128K and integrates improved multilingual capabilities. The model is significantly smarter than its predecessor, scoring 82% on Measuring Massive Multitask Language Understanding (MMLU) compared to GPT-3.5 Turbo's 70%. Additionally, it is more than 60% cheaper.

The new model also includes safety features by default and expanded data residency and service availability. Performance upgrades have been made to Microsoft Azure OpenAI Service as well.

Four in ten workers who use ChatGPT, a popular application of OpenAI's technology, reported that it helped them secure a raise. Three in ten workers said it aided with their promotion. GPT-4o mini is expected to further enhance productivity for these users and potentially attract new ones.

OpenAI has been at the forefront of the AI race since its inception, with ChatGPT attracting one million users shortly after its release in November 2022. However, previous versions of ChatGPT had limited access to data analysis, file uploads, vision, web browsing, and custom GPTs.

The new model's advanced data analysis capabilities allow it to create an interactive table from analyzed data for users to expand on. It continuously monitors the data and provides real-time updates on trends and alerts users when changes occur.

Users can also create custom GPTs tailored for specific tasks or topics by combining instructions, knowledge, and capabilities. ChatGPT's real-time feedback in response to audio inputs is now available in as little as 232 milliseconds, with an average of 320 milliseconds.

OpenAI's mission is to make AI more broadly accessible to people. By releasing a more affordable and capable model like GPT-4o mini, they are taking a significant step towards achieving this goal.


Article Image
Article Image
Article Image
Article Image
Article Image
Article Image

Confidence

91%

Doubts
  • Are there any potential downsides or limitations to using Azure AI for GPT-4o mini?
  • Is the performance improvement of GPT-4o mini significant enough to justify the hype?

Sources

99%

The overall score is a weighted number that takes into account conflict of interest, bias, deception and other practices that undermine the credibility of the source. It is calculated as:
(Site Conflicts Of Interest + Author Conflicts Of Interest) / 2.0 * 0.2 + ArticleBiasScore * 0.20 + UniquePointsScore * 0.05 + DeceptionScore * 0.20 + ReadabilityScore * 0.05 + FallacyScore * 0.20

A score that takes into consideration the content for flow, interruptions with ads, and overt search engine optimization techniques that makes the content hard to understand

OpenAI’s fastest model, GPT-4o mini is now available on Azure AI

Azure by Microsoft Asha Sharma Thursday, 18 July 2024 20:00
  • Unique Points
    • OpenAI has released GPT-4o mini, a more cost-effective and smarter model than GPT-3.5 Turbo.
    • Azure AI provides data residency in all 27 regions and worldwide pay-as-you-go availability for GPT-4o mini.
    • Over 53,000 customers are using Azure AI to deliver innovative experiences at scale, with more than 50% of Fortune 500 companies building applications with Azure OpenAI Service.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    None Found At Time Of Publication
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication

93%

The overall score is a weighted number that takes into account conflict of interest, bias, deception and other practices that undermine the credibility of the source. It is calculated as:
(Site Conflicts Of Interest + Author Conflicts Of Interest) / 2.0 * 0.2 + ArticleBiasScore * 0.20 + UniquePointsScore * 0.05 + DeceptionScore * 0.20 + ReadabilityScore * 0.05 + FallacyScore * 0.20

A score that takes into consideration the content for flow, interruptions with ads, and overt search engine optimization techniques that makes the content hard to understand

OpenAI is releasing a cheaper, smarter model

The Verge Kylie Robison Thursday, 18 July 2024 15:14
  • Unique Points
    • OpenAI is releasing a new, cheaper model called GPT-4o Mini for developers.
    • The goal is to provide something lightweight and inexpensive for developers to create apps and tools they couldn’t afford with larger models like GPT-4.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (80%)
    The article contains selective reporting as it only mentions the scores of OpenAI's model and its competitors without providing context about how these scores were obtained or comparing them fairly. The author also makes editorializing statements such as 'it's worth noting that researchers are wary of benchmark tests like the MMLU,' which goes beyond reporting facts.
    • It achieved an 82 percent score on the Measuring Massive Multitask Language Understanding (MMLU)
    • Google claims Gemini Ultra to have the highest-ever score of 90 percent
    • The new, lightweight model will also support text and vision in the API
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication

99%

The overall score is a weighted number that takes into account conflict of interest, bias, deception and other practices that undermine the credibility of the source. It is calculated as:
(Site Conflicts Of Interest + Author Conflicts Of Interest) / 2.0 * 0.2 + ArticleBiasScore * 0.20 + UniquePointsScore * 0.05 + DeceptionScore * 0.20 + ReadabilityScore * 0.05 + FallacyScore * 0.20

A score that takes into consideration the content for flow, interruptions with ads, and overt search engine optimization techniques that makes the content hard to understand

How ChatGPT's latest version can make you more productive at work

Business Insider Ana Altchek Friday, 19 July 2024 19:40
  • Unique Points
    • Four in ten workers who use ChatGPT reported it helped them secure a raise.
    • Three in ten workers who use ChatGPT reported it aided with a promotion.
    • OpenAI’s new flagship AI model, GPT-4o, has improved capabilities in data analysis and can reason across audio, vision, and text in real time.
    • GPT-4o creates an interactive table from analyzed data for users to expand on.
    • GPT-4o continuously monitors data to provide real-time updates on trends and alert users when changes occur.
    • Users can create custom GPTs tailored for specific tasks or topics by combining instructions, knowledge, and capabilities.
    • ChatGPT can give real-time feedback in response to audio inputs in as little as 232 milliseconds, with an average of 320 milliseconds.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    None Found At Time Of Publication
  • Fallacies (95%)
    The author makes several claims about the productivity benefits of ChatGPT and its new version GPT-4o. While these statements are generally true based on the information provided in the article, there is an appeal to authority fallacy when she refers to OpenAI's CEO calling earlier versions of ChatGPT a 'barely useful cellphone.' This statement is used as evidence that earlier versions were limited, but it does not directly support the author's claims about GPT-4o. Additionally, there are some instances of inflammatory rhetoric when the author describes OpenAI's voice assistant tool being compared to Scarlett Johansson's voice in 'Her.' However, these fallacies do not significantly impact the overall quality of the article and do not detract from its informational value. The article provides clear examples and explanations of how GPT-4o can help make a workday more productive, including advanced data analysis capabilities, custom GPTs for specific tasks, and real-time feedback.
    • ] OpenAI CEO has referred to early versions of the technology as a 'barely useful cellphone.'[
    • Instead of spending 20 minutes re-reading a cover letter or staying stuck on one part of a math equation, ChatGPT can help you on the spot and work through your questions as you go along.
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication