The new GPT-6 Astra is poised to make significant waves in the AI market, showcasing overwhelming performance and efficiency in terminal benchmarks.
Want the next one from this channel too?
GPT-6 Astra was unveiled a few hours ago. It gives the impression of being hastily released to counter Fable 5.1, which launched yesterday. Currently, Astra is available to a limited number of users.
GPT-6 Astra claims to deliver nearly three times the performance at the same price, based on Terminal Bench SAS 0.1. This performance difference is noticeable even for general users.
It's no longer just about higher performance; cost and speed must now be considered when selecting a model. Models that combine cost-effectiveness and performance will be key elements of frontier models.
It is anticipated that once Astra is publicly released, it will be the most affordable yet best-performing frontier model. This is attributed to Astra being released with a very high level of optimization.
There have been significant advancements in technology that accurately recognizes the location and space of major interfaces and takes action accordingly. From this perspective, GPT is expected to move much further ahead in computer automation. This technology will be crucial even in areas where browser automation was difficult, such as KakaoTalk or screen capture.
The previous model, Sol, did not show superior performance compared to Fable. However, GPT-6 Astra has significantly improved its physics engine and overall performance, increasing efficiency even in creating complex applications like traffic simulators. Benchmark results directly translate into crucial work efficiency here.
Achieving high scores in screen reading benchmarks is essential for accurately inputting data at the correct location and deriving precise values through accurate analysis. This contributes to user satisfaction by improving accurate location recognition and analysis capabilities in tasks like form filling.
GPT recorded high scores in the screen-related computer use benchmark this time. This leads to improved Power BI automation and enhanced mouse control capabilities, resulting in higher efficiency in tasks such as CAD modeling and document organization. GPT is actively emphasizing this aspect.
Along with GPT-6 Astra, the cex harness update has significantly improved computer usage speed. This engine update is expected to lead to higher benchmark scores and practical reductions in work time.
GPT-6 Astra can achieve the same results in creative workflows using 20% fewer tokens than before. This signifies high efficiency in tasks such as Unreal Engine work or 3D model generation.
Astra emphasizes a method of asking users questions and improving results based on their answers when there are elements needed for improvement that the user may not be aware of. This contrasts with the previous model, Sol, which tended to handle things autonomously even with insufficient information. This demonstrates its ability to improve results through collaboration.
GPT-6 Astra began rolling out to some users today. It is expected to expand to Plus, Pro, Business, and Enterprise subscribers within a few days. Astra will be included in the basic subscription model at no additional cost.
Answers come from the transcript, with the exact spot cited.
Want the next one from this channel too?
When 코드팩토리 publishes, we'll write it up like the one you just read and email it to you.
{channel} published 6 in the last 7 days.
We skip Shorts. You can unfollow any time.
